Education-line Home Page

The Costs and Benefits of Lifelong Learning:

The Case of The Netherlands

Marko J. van Leeuwen

Foundation for Economic Research (SEO)
University of Amsterdam, the Netherlands

Paper Presented at the European Conference on Educational Research, Lahti, Finland 22 - 25 September 1999


The paper deals with costs and benefits related to vocational education and on-the-job training. After a literature overview on lifelong learning, explanations for failures on the market of vocational education are explored. Next the paper describes the actors and the costs and benefits items relevant in the vocational education market.

For the calculation of costs and benefits of on-the-job training on meso and macro level a model is developed. Model parameters are estimated using information from a query under employers and employees in the Netherlands. Exogenous model variables are taken from the query as well as from several statistical sources.

The model is used for running a baseline scenario and several policy scenario’s. The policy scenario’s describe proposed policy measures for the stimulation of lifelong learning in the Netherlands. The model calculates detailed costs and benefits for actors on the vocational education market and the macro-economic consequences.


Research in the field of education is mainly focused on pedagogical, sociological and organisational aspects of learning and training. Still, the acceptability of new policies depends also on the money available and costs. A rightful question also in education is how the scarce financial resources are most efficiently distributed, i.e. taking into account both costs and benefits. In this paper the costs and benefits of lifelong learning are evaluated.

In December 1997 the Dutch government launched the National action plan ‘A Lifelong Learning’ as part of the National Knowledge debate (Projectteam, 1998). SEO was asked by the ministry of Education, Culture and Sciences to make a study of the economic effects of (fiscal policy measures for the stimulation of) lifelong learning. The study comprises three parts: 1) a literature search after elements determining failures on the market of vocational education and training; 2) a query among employers and employees focusing on on-the-job training and 3) construction of a calculation model for the costs and benefits of lifelong learning.

The potential failures on the market of vocational education and training are briefly described in section 2. In order to be able to evaluate the costs and benefits of the introduction of policy measures aimed at stimulation of lifelong learning a calculation model is developed. In the model the behavioural effects on micro-level are translated into sector, meso and macro-economic level. The model is described in section 4. The model parameters are estimated using information from a query under employers and employees in two economic branches in the Netherlands. Exogenous model variables are derived from the query as well as from several statistical sources. A description of the query and a synopsis of the results are found in section 3. The calculation model is used for running a baseline projection and several policy scenario’s. The results are given and compared in section 5.

Failures on the Market of Vocational Education

The starting point of the analysis is Beckers’ Human capital approach which suggests that only general training is transferable from one job to another. Specific training can only be applied in the agency in which the skills are acquired. Therefore, under-investment in vocational education and on-the-job training is likely. In the literature various types of market failure are described. They can also serves as explanations for the, from and economic point-of-view, suboptimal allocation of resources for lifelong learning. First of all, there can be an information problem. It is not always exactly clear what is learned during a training and therefore what will be the return, both for the trainee (higher wage, improved employability) and the firm (improved productivity). Another prominent reason for mismatches on the market of permanent education is the poaching problem. Firms face the risk of other employers ‘poaching’ trained workers, or of workers bidding up wages as a consequence of acquiring a higher qualification. The, in training investing, firm risks ending-up with less than the full range of the expected returns of training. On the other hand individuals may withhold from (investing in) training if gains in qualifications and competence are not adequately recognised, or if there is uncertainty about the returns as was mentioned before. This last issue is especially important if individuals have little financial resources and are facing a liquidity constraint. Which means that if future returns of investing in training are uncertain, they are not suited to serve a collateral for this investment. Borrowing on the money or capital market is than either impossible or at least very expensive. The information problem, the liquidity constraint and the uncertainty result in less than optimal investment in training.

Other types of market failures, here to be mentioned but not spelled-out, are economies of scale, poor basic skills of (certain parts of) the population, external effects, the social security system (amongst others minimum wage legislation), (too) narrow wage differentials and high unemployment rates.

Several approaches are suggested for addressing the problems that arise from the various types of market failures. They can be divided into two groups: financing models (single employer financing, self-financing, drawing rights, vouchers, actions and para fiscal fund) and instruments (fiscal instruments, training-subsidies, information, certification of diploma’s, accreditation of training institutes, apprenticeship contracts, differentiation of etc.). The paper focuses on fiscal measures that are considered in preparation of the aforementioned National action plan: a Life Long Learning. The measures considered are described in section 4.

The cost-effectiveness of a policy measure depends also on the level of aggregation and the point of view. As will be shown in section 4 a more cost-effective policy measure on a macro level can have large impact on the distribution of costs and benefits between actors. A distinction can be made of who is affected (positively of negatively) by the policy measure. In describing the costs and benefits we specifically look at three types of actors: employees, firms (employers) and the government (not as employer).

Vocational Education in the Netherlands

On average the population in Netherlands is highly educated. An international comparison shows that the Dutch education system belongs to the world top-10 (The Economist, 1997). But, as the OECD (1996, pp.122) recognises: ‘Lifelong learning presumes continuity between initial education and training, and the organised learning experiences that take place thereafter, during working life. What is sometimes overlooked is that continuity is absolutely essential throughout initial education and training.’ Therefore, in the next subsection a brief description is given of the vocational training activities in the Netherlands. In section 3.2 and 3.3 the results of the queries regarding training in two economic branches are described.

Employer Sponsored Training

In 1993 45% of all enterprises in the private sector with five or more employees did some kind of training effort. One out of four employees in the Netherlands followed that year one or more internal or external courses, while approximately 12% underwent some training in the work situation. The total number of attended internal and external courses was 1.2 million (an increase of 5% per year since 1990), with an average duration of six days. About 60% of the man-days in training was given in working time. The economic branches ‘financial intermediation’, ‘transport & communication’ and ‘public services’ show a much more than average participation in training, while the sectors ‘agriculture’, ‘construction’ and ‘hotels & restaurants’ show relatively low participation.

Dutch enterprises spent in 1993 about 3.5 billion guilders on internal and external training (including cost of lost working time, fees of training institutes, compensation to workers and other costs). This is approximately 1.7% of total labour costs (2.3% if only enterprises with courses are considered), which is on average 990 guilders per employee (or 1.350 guilders if only enterprises with courses are considered). Per attended course the average costs where 2.820 guilders.

Query Among Employers and Employees

The information on employment, wages, taxes and on-the-job training in the Netherlands from official sources (CBS, branch organisations) has certain drawbacks. First of all, it dates back to 1993 and is therefore somewhat outdated. Another drawback is that it does not give any detail on the reasons for following training (or not). Finally, the aim of the model is to evaluate the effects of new policy measures, which cannot be measured and quantified by using existing information. Therefore, four queries were held by Intomart among employers and employees in two economic branches. For each query 175 persons were approached of which approximately 120 are interviewed; adding-up to approximately 480 interviews. Through the queries it is possible to determine the amount of training consumed, the costs of training, who pays for the training (in money and time) and the effects of (additional) training in terms of productivity, employability, job security and wage levels. Furthermore the respondents are asked to indicate the attractiveness of several (fiscal) policy measures aimed at stimulation of training. The information from the queries is used to estimate the model parameters.

The results from the queries show amongst others that in the Netherlands both the need and the willingness to participate in on-the-job training is high. Especially learning by doing, learning from colleagues and combinations of learning and working are popular. The direct costs of on-the-job training are almost completely paid by the firms. Therefore fiscal stimulation measures are most effective if directed towards them.

The driving force behind training is for employers increasing the skills of their employees. Increasing employability is the key factor behind training for employees. They want training to increase job security, increase job mobility and to keep their skills up-to-date.

Women appreciate training better than men, while employers choose men above women when training is offered. Employers also prefer to train full-time staff above part-timers. Both effects are correlated since women in the Netherlands work more often part-time than men. Furthermore, participation in training activities is sector specific and is higher in bigger firms and higher for higher educated employees. Finally, older workers (age: 45+) and worker staying longer with the same firm are trained significantly less than younger workers (15-44) en new employees; again these factors are strongly correlated. In order to increase the employability of women, part-timers and senior staff, special attention to those groups is recommended.

Conjunct Analysis

Since, we deal with new policy measures it is impossible to measure the effects directly. The complexity of the problem requires a special approach. In order to discover the preferences of employers and employees concerning fiscal stimulation measures of on-the-job training we use ‘conjunct analysis’ of ‘vignette analysis’. In this method respondents are shown several training profiles (‘vignettes’), with each profile consisting of a set of characteristics or attributes of a certain type of training within a specific setting. The attributes are chosen carefully, each of them representing a dimension which is important to the decision making process of the respondent. The dimensions or attributes selected are: characteristics of the training (type, time, duration, and price/costs), the fiscal measure considered and a set of remarks concerning the results of training and other circumstances. Figure 1 gives an example of an ‘employee vignette’ in the sector commercial services. Through varying the attributes a large number of different profiles or vignettes are created.

vignette 7

type of training special external practise training

time and duration one week during working hours

total price ƒ20.000,— ; fully paid by employer

subsidy for employer none

subsidy for employee 100% subsidy for paying wages during training

remarks employability increases after the training;

employee signs a training contract

Figure 1. Employees vignette; commercial services. Source: SEO/Intomart

The respondents are asked to score to each of the eight vignettes selected for them on a ten point scale based on 1) the attractiveness of the vignette, and 2) the chance of accepting the training in given circumstances. The scores are analysed using both Ordinary Least Squares and Ordered Probit, with the vignette attributes and other characteristics of the respondents (as taken from the rest of the query) as explanatory variables. The advantage of this method is that it is far better in finding answers to revealed preferences of respondents than any direct way of asking. Furthermore, the method makes it possible to detect and quantify the effects of new policy measures within a realistic setting. The results can easily be translated into model parameters, measuring the sensitivity of respondents (employers or employees) with respect to the attributes in the vignette.

Analysis of the vignettes shows that both employers and employees with a schooling past have a more positive attitude towards future training, than inexperienced ones. This is especially true for former external training. Therefore, training is a positive stimulus for lifelong learning. The important policy advise here is that once the flywheel is at steam, for example generated by a government measure tackling one of the market failures, the system of learning will continue more or less automatically.

Another observation is that compulsory training measures are more effective in terms of participation rates, but at the cost of lower valuation of the measures. Which effect is strongest in the longer run is unclear. Compulsory training measures are therefore not recommended.

The employer and the employee both try to persuade the ‘other party’ to invest money and time in training. In the Dutch situation employers are the main financial contributors of on-the-job training. Fiscal measures are therefore most effective if directed towards the employers. Also employees react more positive on measures on the side of the employers than those directed towards themselves. The willingness for training in own time is low for both employers and employees. An effective policy measures is one that compensates the party that invests time in the training, for example an employers’ compensation for forgone hours due to absenteeism of the employee. Also stimulation of more flexible training schemes and reciprocal courses are most effective.

The SEO Calculation Model Lifelong Learning

SEO has developed a model for the calculation of costs and benefits on sector, meso and macro level, of policy measures aimed at stimulation of on-the-job training. The model focuses on the working population in the Netherlands and distinguishes between nine economic branches, two types of training (internal and external) and 216 types of employees (amongst others age, gender, education level). Furthermore, the model distinguishes the schooling related costs and benefits, distinguished by type of actor, as described in figure 2. The base year of the model is 1996; the year for which at the time of construction of the model in 1997 most detailed information as described above was available.




higher wage after training costs of training born by employee
extended working lifetime crowding out effects
additional labour supply loss of spare time and vacation
labour satisfaction, increased status, learning is ‘fun’  
job security, labour market opportunities, employability  


increased productivity (after training) loss of productivity (during training)
stronger commitment to firm costs of absenteeism (during training)
reduction of labour market bottlenecks higher wages
less deterioration of human capital costs of training born by employer
better educated and more productive population subsidies paid
additional labour taxes and social security premiums tax exemptions given
reduction of allowances (sickness, unemployment, etc.) social security payments

Figure 2. Inventory of costs and benefits of on-the-job training, by actor. Source: SEO

Model Overview

An overview of the model is given in Figure 3 on the next page. The behavioural relations in the model (endogenous variables) are indicated by boxes with round corners, while boxes with squared corners indicate model variables determined (exogenously) outside the model and definition equations. Looking at the figure three broad lines of relations can be distinguished. First, there are equations concerning employees; there personal characteristics (gender, age) and labour characteristics (sector, wage, type of contract, working hours, taxes, social security). Second, the figure displays a block of equations on training and education. (participation rates, types of training, costs of training, contributions in time and money). Finally, there are equations in the model describing the effects of training for employees (making promotion, higher wage, increasing employability, etc.), employers (changes in productivity during and after training, flexibilisation, employability) and the government (taxes, subsidies).

Fiscal Policy Measures Considered

In the query and the model simulations amongst others seven different fiscal policy measures are considered. They are described below together with the sensitivity as measured in the OLS and Ordered Probit analysis.

Refunding of (50 to 100%) of the costs of training paid by employees. Results in a 24 to 51% increase in the willingness of employers and employees to participate in training in the commercial services sector. In the construction sector the effect is in the range of 55 to 116%.

Refunding of 100% of the costs of training paid by employers. This measure has no effect on training activity in the construction sector, probably due to the already existing schooling fund. In the services sector the willingness to participate in training increases by 248%.

Refunding of 100% of the costs of loss of production due to absenteeism. In the construction sector and the services sector the willingness to participate in training increases with 56% and 256%.

Income tax reduction if employee sacrifices free time for training (ADV). In the services sector this measure has only a small positive effect on the amount of training. In the construction sector the effect is even slightly negative.

Subsidy for 50% of the direct costs of training for both the employer and the employees. In the construction sector and the services sector the willingness to participate in training increases with 67% and 138%.

Combination of employee sacrificing free time (ADV) with a 100% refunding of the direct cost of the training by the government. The effect on the willingness to participate is 130% in the construction sector and neglectible in the services sector.

Refunding of 50% of costs of absenteeism and an investment premium of 50% of the direct costs of the training. The effect on the willingness to participate is 47% in the construction sector and 300% in the services sector.



Figure 3. Schedule SEO Calculation Model Lifelong Learning. Source: SEO.

Scenario Results

The calculation model is used to run a baseline projection and six policy scenario’s, that are (combinations of) the policy measures described above. In every model run detailed costs and benefits are calculated for the three types of actors on the vocational education market and the macro-economic consequences in terms of number of jobs created, productivity increase and budgetary results. It is also possible to show results on more detailed level: sector, type of employee or combinations of these.

Table 1

Key indicators of on-the-job training in the Netherlands: baseline projection (1996), by sector

gross wage bill fl billion value added fl billion employed persons x1000 days of internal training days of external training % training in time of employer Training costs fl million absenteeism fl million
Construction 14.2 21.2 227 312 495 57% 176 67
commercial services 11.2 25.8 168 45 885 55% 545 412
other construction 9.7 11.6 163 264 450 54% 144 55
other services 38.9 139.5 600 172 2,586 54% 1,345 960
agriculture 12.5 19.2 218 237 668 58% 162 64
trade and transport 63.3 126.1 1,405 1,276 4,617 57% 1.313 716
industry 58.0 152.9 1,030 1,240 3,456 57% 1,177 582
health care 27.9 47.2 788 446 3.597 56% 965 583
government 61.3 91.4 1,123 625 4.327 58% 1,289 666
total 296.8 634.9 5,726 4,615 21,081 57% 7,117 4,105

Source: SEO

The baseline projection is fine-tuned, to match reality in the base year as close as possible, by changing the exogenous parameters of the model. Table 1 gives the key indicators per sector of economic performance (gross wage, value added, employment) and training (number of training days, training costs) as calculated by the model for the base year 1996.

Table 2 shows the results of six policy scenario’s in deviation from the baseline projection. From an macro-economic point of view the introduction of a 50% employer subsidy (scenario a) is most profitable. The total net effect is 315 million guilders. A disadvantage of the scenario is that it generates a rather large transfers of income between the three types of actors. If the employer subsidy is further increased to 100% of total costs of training (scenario b), the additional effectiveness of the measure is rapidly decreasing, and becomes extremely negative macro-economically and for the government. A less profitable, but much more balanced scenario, is the lower profit tax of ƒ250 per trained employee; 228 million guilders is generated at the expense of ‘only’ 24 million guilders net investment by the government.

Table 2

Summery of the overall results of the baseline projection and the policy scenario’s, deviations from the baseline projection in millions of guilders






a) 50% employers subsidy





b) 100% employers subsidy





c) employees trained in spare time (ADV)





d) ADV + employees pay extra costs of training





e) lower profit tax of ƒ250 per trained employee





f) ADV+ 25% subsidy of training costs





Source: SEO

Final Remarks

With the calculation model described in this paper it is possible to make estimates of the costs and benefits of policy measures aimed at stimulation of vocational education in the context of lifelong learning. Although the scenario’s described are merely illustrations of the working of the model, they show, beyond any doubt, that the differences in cost-effectiveness of this kind of policy measures can be large and the results can be .

In order to improve the quality of the output of the scenario’s further research after the model parameters is needed. For example through additional queries amongst employees and employers in other sectors. The Dutch Central Bureau of Statistics is currently updating the 1993 survey of employer sponsored training. This will increase the knowledge of on-the-job training in the Netherlands and will be most useful for improvement of the model. The approach can also be extended to other countries if sufficient data are available or collected.


Barron, J.M., M.C. Berger and D.A. Black (1999), Do workers pay for on-the-job training?, the Journal of Human Resources, vol. XXXIV, no.2, pp.235-252.

Becker, G.S. (1964), Human Capital: a theoretical and empirical analysis with special reference to education, Cambridge University Press, Cambridge.

Beek, K.W.H. van, C.C. Koopmans and B.M.S. van Praag (1997), Shopping at the Labour Market, a Real Tale of Fiction, in: European Economic Review no.41, pp.295-317.

CBS (1995), Bedrijfsopleidingen 1993; particuliere sector [Employer sponsored training in the Netherlands in 1993; the private sector], the Netherlands Central Bureau of Statistics, Voorburg/Heerlen.

The Economist (1997), Education and the wealth of nations, 29-3-1997.

Leeuwen, M.J. van, I. Overtoom en B.M.S. van Praag (1997), De kosten-effectiviteit van een leven lang leren [the cost-effectiveness of lifelong learning], SDU publishers, the Hague.

OECD (1996), Life-long learning for all, OECD report, Paris.

Oosterbeek, H. (1996), Financing lifelong learning, Mimeo, University of Amsterdam

Projectteam ‘Een Leven lang leren’ (1998), Nationaal actieprogramma een leven lang leren [National action plan: a Lifelong Learning], Ministry of Education, Culture and Sciences, Zoetermeer.

This document was added to the Education-line database 23 September 1999