The work approach in norway aims, measures and results Torkel Bjørnskau Espen Dahl

Download 91.69 Kb.
Date conversion04.02.2017
Size91.69 Kb.
the work approach in norway – aims, measures
and results

Torkel Bjørnskau

Espen Dahl

Jens B. Grøgaard

Fafo Institute for Applied Social Science

P.O. Box 2947 Tøyen, N-0608 Oslo, Norway


Norway, Sweden, Denmark and Finland have all had relatively generous welfare regimes, based on universal rights, collectively referred to as the Scandinavian model. Although there are differences between these countries, which are perhaps increasing, the Scandinavian model is distinguished by the automatic granting of relatively large compensatory income from the state to those who lose their jobs, get sick, disabled or are otherwise excluded from the work force. According to Esping-Anderson (1990), the Scandinavian model or social democratic model can be distinguished from:

  • a conservative model or continental model which is typical for Germany, France and other European countries where the welfare system is not based on universal rights, but on occupational status and negotiated rights and

  • a liberal model, typical of Great Britain and USA where welfare provisions are typically based on individual insurance schemes and where the public welfare provision is limited and based on means testing.

Recently, the welfare state has been subject to serious attacks, mainly for fiscal reasons. The fiscal problem is both the problem of financing today's generous arrangements as well as the problem of financing pensions and care of an increasing number of elderly people which most western countries will meet during the next century. Many European countries, which have steadily reduced the age of retirement, both as a measure to increase welfare and as a measure to reduce unemployment, are now rethinking and reversing this trend. In Germany, there were strikes and demonstrations earlier this year, a reaction to precisely such an attempt at increasing the age of retirement. Sweden has radically reduced the level of compensation in most of its welfare state arrangements over the last few years due to fiscal problems and substantial public debt. These reductions have triggered widespread protests, with pensioners demonstrating in the streets, and increased the number of poor people in Sweden.1.

The Swedish experience illustrates the risk involved with a welfare state providing generous benefits based on universal rights. The problem is quite simply that a generous welfare state can only survive as long as the number of social security beneficiaries are limited. When the number of beneficiaries increases beyond some critical threshold the fiscal strain upon the welfare state enforces a reduction in the standards. This is what happened in Sweden. An even though many, including the former Minister of Finance, Feldt, saw the writing on the wall, many did not. Perhaps also the system was too rigid, and consequently the problems became very large.
In order to avoid the kind of problems experienced in our neighbour country the Norwegian Government has adopted the so-called "work approach". According to the White Paper on Welfare Policy passed by the Government in 1995, this strategy entails that work must be the first preference, and that policy measures and programmes should facilitate labour market participation among the young, sick, disabled, single supporters and elderly.
The work approach has two key aims, according to the White Paper on Welfare Policy. First, the work approach is intended to reduce the need for redistribution from the occupationally active population to social security beneficiaries. Second, the work approach is intended as a means of enhancing individual well-being, taking as given that remunerated work increases quality of life, both economically and otherwise.
The work approach consists of numerous measures, including some designed to provide incentives to work (or education) and others making it more costly to rely on social welfare. Among the former are an array of measures aimed at reducing sick absenteeism among the long-term sick. Norway has a very generous sick leave compensation scheme, providing 100% compensation of pensionable income from the first day of sickness. Over the last two years, sick absenteeism has increased, raising the question of whether this is a logical consequence of the work approach, or an exploitation of a generous sick-leave (Pedersen 1996).2
The measures adopted to reduce sick absenteeism include programmes of work participation during sickness3, follow-up programmes for long-term absentees, the implementation of internal work controls schemes etc. The results of an evaluation of one such follow-up programme will be presented below.
In addition to implementing programmes to reduce the level of sick absenteeism, i.e. to keep people in paid work, the work approach also includes numerous measures to move people from welfare dependency to paid work. Different work-training programmes have been adopted, especially directed at young people. Two such work-training programmes have been evaluated by Fafo, and the results are also presented below.

can long-term sick absentees be talked back to work?

Is it possible, by fairly simple and inexpensive means, to enhance return to work among long-term sick absentees, and can the local National Insurance Office play a significant role in this process? The aim of the project4 described in this section was to shed light on this question by evaluating an intervention carried out by three local National Insurance Offices in a county in Norway. In this section we describe the background, the organisation and the design of the project, present the most important results, and discuss their implications. The intervention was designed as a randomised controlled trial with the ultimate aim of reducing both the number of sickness days and the sickness benefit costs. The intervention was a personal talk between the sick absentee and the social insurance officer which took place after four, seven and ten weeks.


The work approach of the Norwegian Government is an important part of the conceptual background for this project, providing the ideological climate which led to its initiation and enhanced its financing. As part of implementing the work approach the local National Insurance Offices were assigned more responsibility and extended powers, especially versus the health care system and the physicians. For example, the local National Insurance Offices were given the authority to overrule the decisions made by certified physicians. The local National Insurance Offices were also encouraged to be more proactive towards the health services, the employers, other public agencies and towards the long-term sick themselves.

In addition to the above-mentioned factors, the regional National Insurance Office experienced some, for them more unpleasant problems, namely that the county in which this project was initiated was above the national average with respect to sickness benefit costs, number of sickness days, and the number of long-term sick. These problems prompted the question of whether the local National Insurance Offices could implement some measures to improve this position. They reasoned, rightly or wrongly, that being on sick leave for weeks or months may have a number of unfortunate consequences. It may undermine the motivation to return to work: it may create passivity and social isolation; and it may generate a feeling of being worthless and expendable. The regional National Insurance Office also suspected that too many people did not get the medical treatment they needed, or got it too late.
There are set procedures on how to follow up long-term sick after eight weeks. Physicians are obliged to fill in a special form which among other things, sets out their assessments regarding future treatment and assistance, both medical and vocational. However, many local National Insurance Offices do not adhere strictly to these procedure. Thus, in many offices, it should be possible to adjust practice in such a way as to improve the follow-up routines for the long-term sick.
The regional National Insurance Office held the view that current practice suffered from two serious weaknesses, which could be modified. The first was that contact between the sick and the local National Insurance Office was established too late; the second was that the relationship between the patient and the local National Insurance Office was too impersonal. Thus, it was decided that the project should be aimed at eliminating these two deficiencies.
In order to provide reliable data, it was decided that the project design should be as close to a clinical controlled trial as possible. An additional point in this respect was that controlled trial designs are very rare in this field of research in Norway (Tellnes et al. 1992). However, scattered evidence (based on controlled and semi-controlled trials) exists of interventions which have had some limited positive effects on sickness duration and the sickness benefits (Melsom 1986, Melsom and Noriek 1988).

The Research Design

The project was designed as a randomised controlled trial (Pocock 1983). Individuals had to meet four criteria in order to be included in the trial. They had to be:

  • below 65 years of age;

  • employed (self-employed were excluded);

  • certified 100% sick; and

  • diagnosed as having either musculo-skeletal diseases or nervous ailments.

There are several reasons why these two diagnostic groups were selected. First, these groups often do not have a precise and unequivocal biomedical definition. That is why they are often called diffuse ailments. This means, among other things, that the sickness duration may vary widely from person to person, and that the degree of working capacity can be difficult to determine. In addition, the health care system and other agencies may have little to offer to these patients other than issuing a "wait and see" prescription. Thus people with ailments like these may be socially vulnerable and may be receptive to information and encouragement.

The project was organised as a collaboration between researchers, social insurance officers from the local National Insurance Office, and representatives from the regional National Insurance Office. It was of crucial importance to establish a consensus among the collaborators as to how both the selection of individuals and talk programme should be carried out. Thus, much time was devoted to developing a common understanding of aims and procedures.
The Insurance Office at the local level was given the responsibility of implementing the intervention and collecting the data. The regional level operated as an initiator and as a catalyst, while Fafo Research Institute was responsible for the scientific part of the project. No researcher participated in the talks, in the selection of individuals, or in the collection of data.
It was decided that each local National Insurance Office should include about 70 sick absentees in the trial group, and a similar number in the control group, randomly selected. Persons in the trial group were included successively, in order to spread the work load over time; the project took place within the usual day-to-day routines of the offices.
The sick absentees in both groups were identified after a three-week sickness episode5. Those who were selected to join the trial group were invited by mail to come to the local National Insurance Office for a personal talk with a social insurance officer. In effect, the sick absentees had little choice other than to accept the invitation, but some talks were undertaken over the telephone. If the physician reported the person fit between the third and the fourth week, the person was removed from the trial, regardless of group to which she or he belonged. A total of 202 people were included in the trial group, and 205 were included in the control group.
For economic reasons, the number of talks with each person in the trial group was limited to a maximum of three, held after four, seven and 11 weeks. A "talk form" was constructed, very much like an interview guide, in order to structure the talks. This form contained a progression from one talk to the next so that the talk programme should not be too boring or repetitious for the sick absent. The more specific aims of the talks were to map the situations of the participating individuals, to make them conscious of their circumstances, stressing their responsibility, and, if necessary, to motivate them to do something about it. Topics included the background for the invitation, occupational conditions, the participant's health situation, medical treatment, and what to do next.

Data and Methods

Our dependent variables were number of days receiving sickness benefit, and gross sickness benefits. For individuals in both groups we collected information about gender, age and diagnosis. Thus, we have the possibility to examine whether the outcomes vary according to these background variables and to control for variation in them. The status of cessation has three values, i.e. reported off the sick list, maximum sick absence period (one year), and "ceased for other reasons".

The data sources are the personal files located in the respective local National Insurance Offices. The local project collaborators transferred information from these files to registration forms designed for this purpose by the working group. These forms were made anonymous and mailed to Fafo Research Institute where they were coded as processed. The data from the people in the control group were processed in exactly the same way.
The data analysis was conducted by means of anova, survival analysis (Caplan Meier estimation) and Cox-regression. The two latter methods are used to gain further insight into the duration patterns of the two groups and to handle the problem of censoring, i.e. to assign values to those who exceeded one year sick leave and those who ceased for other/unknown reasons.


Table 1 shows the main outcomes for the two groups. The effect of the trial is indicated by the differences between the control group and the trial group.

Table 1 Status of Cessation, Mean Duration of Sickness Days and Sickness Benefits by Trial Group Control Group and Total.

Trial group

Control group

Difference (C – T)


Duration in days*





Standard deviation




Sickness benefit costs

NOK (Norwegian krone)**





Standard deviation




Status of cessation (n.s.).

Reported fit %





Maximum period %





Other reasons for cessation %





* P-value .06

** P-value .30

n.s. = not significant

Table 1 shows that on average, the trial group had 19 sickness days less than the control group, and close to 4360 NOK lower sickness benefit. However, it is only the difference in days of duration which approaches statistical significance. We observe as indicated by the standard deviations, that there are large dispersions within the two groups both for sickness days and for sickness benefits.

We may also note that the differences between states of cessation are small and insignificant. This means that although the intervention seems to have had the desired effect on the duration of the sickness episode, it had little effect on the probability of staying sick the whole one year period.
This pattern repeats itself in different subgroups (results not shown here). The differences are, however, more pronounced among men and women, among the elderly than the young, and among those with musculo-sceletal illnesses than those with nervous ailments.
A general tendency is that the differences are statistically more significant terms for duration of sickness days than for sickness benefits: 19 days reduced sickness absence days compared to only 4356 NOK reduction in sickness benefit costs per person. Why do the two outcome measures diverge to this extent? A plausible explanation is that the likelihood of being reported off the sick list may be higher for those with low incomes than for those with higher incomes. This possibility is examined in Table 2, which shows the sickness benefit provided to each person per day among those who had a sickness absence below and above 150 days of duration.

Table 2 Sickness Benefits Provided Per Day Per Person in the Trial Group and the Control Group According to Duration of the Sickness Episode.


Below 150 days

151 days and more

Trial group




Control group




Difference (T –C)








Table 2 shows clearly that the difference between the two groups in benefits per person per day is much larger, i.e. 35 NOK, among those with longer absence (151 days and more), than among those who had a shorter sickness episode. In this latter group, the difference is negligible, only 2 NOK. Thus, it seems that one important reason for the modest reduction in sickness benefits is the stronger tendency for the low paid to return to work over time in the trial group than in the control group. This is equivalent to say that the intervention is more effective for the "poor" than for the "rich". We will return to further discuss this finding.

Survival Analysis Discussion Section

One problem with the anova analysis is that the technique does not handle censoring, i.e. what to do with cases for which there was incomplete information or who had exceeded the one year limit. To meet this problem, and to give a picture of how the two groups "behave" over time, we have carried out a survival analysis on the duration data. Figure 1 shows the survival curves for the trial group and the control group.

Figure 1 Survival Curves for the Trial Group and Control Group
(Caplan Meier Estimation)

Figure 1 shows that the trial group displayed higher rates of being reported fit than did the control group from early on. However, this tendency diminishes around seven months and becomes almost absent from nine months and on. The difference between the two groups is statistically significant (p-value = 0.03). Similar to the data displayed in Table 1, Figure 1 shows little difference between the two groups with regard to the duration of sickness days is independent of sex, age and diagnostic group. This is evident from a multiple Cox-regression analysis (results not shown here).

Summary and Discussion

Our analysis shows that, at least to some extent, the "talk programme" operated according to intentions. On average, the number of sickness days among those in the trial group is 19 days less than in the control group. This difference is significant, but only just. The sickness benefit costs are reduced accordingly, i.e. by about 4350 NOK, but this outcome is far form significant in statistical terms. The finding that the intervention has some affect in the right direction is in accordance with evidence from previous research in Norway (Melsom 1986, Melsom and Noreik 1988, Pedersen 1993) and from numerous evaluations of the so-called "workfare" projects in the US (Gueron and Pauly 1991).

The peculiar discrepancy between the two outcome measures is worth a comment. There is obviously a selection going on in the trial group which implies that those with lower incomes return faster to work than those who are better off. We do not know if this selection process is unique for this intervention, and we can only speculate about the underlying mechanisms. One interpretation is medical: it is possible that individuals with lower incomes have less serious diseases and lighter work incapacity so that is easier for them to adhere to the wishes and messages from the social insurance officer. Another interpretation is sociological: low income individuals are more apt to regard the social insurance officer as a powerful authority, and therefore feel a stronger obligation to comply with the (perhaps subtle and implicit) messages he or she sends. Put in other words: are the poorer patients more vulnerable and easier to "scare" back to work than the more wealthy? If this latter interpretation is correct, we face a problem. First, it will not be just and fair if the intervention works most effectively upon individuals who are less endowed with material and social resources. Second, we cannot ignore the possibility that these persons may return to work too early, so that in the long run they may have a greater risk of relapse.
The talk programme seems to influence the number of sickness days, but does not reduce the risk of exceeding one full year sick leave; the percentage doing so were the same in both groups. More than one year of sickness absence may indicate a raised risk of a more permanent exclusion from the labour market. Seen in this perspective, it seems that although the talk programme may reduce the number of sickness days, it is not well suited to prevent more permanent absence from the workforce. This result seems reasonable as the forces that drive complete exclusion from working life are probably so strong that they can be only modestly counteracted by interventions like the one studied here.
As indicated by the standard deviations, there are wide variations in sickness days and sickness benefits within both groups. This heterogeneity is to a very limited extent explained by sex, age and broad diagnostic groups, and is one reason why many of the differences do not reach statistical significance. Thus, there must be other sources of this internal dispersion. These sources may, for example, include working conditions, or the degree of severity within more specific and unitary diagnostic groups. In order to make this kind of trials more effective, there is still much to be learned about which target groups one should single out for intervention.
An important question which we have not been able to address in this project is whether an intervention like this exerts only a short-term effect. We do not know whether people in the trial group were certified sick again shortly after they were reported off the sick list, or if they had a higher risk of being certified sick once again later on. Future research should make some effort to address these important questions properly. In Norway, the necessary data exist on files located in the local National Insurance Offices; in the future they should be utilised. Pursuing this point further, we would add that interventions like this talk programme are likely to be effective only as long as there is some slack or potential for improvement in the organisation. Otherwise it is likely that their impact will be small, and once established as a permanent routine, any additional effects are likely to be negligible.

do prepared job-training programmes for
unemployed young people make a difference?

Talcott Parsons stated that the educational revolution is as characteristic of modernity as the industrial and democratic revolutions have been (Toby 1977:190). Certainly to this extent Norway is a nation now facing modernity. The country has experienced a rapid growth in the educational level of the population since World War II, and at present close to 50% of new cohorts are expected to enter various institutions of higher education6.

At the same time, there has been a successive decline in the traditional employment opportunities for adolescents (16-24 years of age) seeking ordinary paid work. Some figures may illustrate the point. In 1975, one third of the young people 17 to 19 years of age reported that their main occupation was ordinary paid work, while six in ten reported that they were pupils (basically at secondary level). Ten years later (1985), the proportion of pupils had increased to 75%, while the proportion of ordinary employees was 17%. In 1990, 86% claimed that they were pupils, while only 8% reported that their main occupation was ordinary paid work (Moen 1991.) At the same time the unemployment level among these teenagers was 18% (Central Bureau of Statistics, Norway)7. As a frame of reference for the research projects described below, the following structural features should be kept in mind:

  • The former labour market for young unskilled assistant workers virtually disappeared during the 1980s. At the same time, the country witnessed a rapid growth in part-time work (at weekends, evenings etc.), which probably represented an Eldorado for young people who combined schooling and employment (Grøgaard 1993).

  • The Norwegian authorities struggled to compensate for this decline in permanent job opportunities through a rapid expansion of the capacity of secondary education in general, and of vocational education in particular. This educational policy was accompanied with active employment measures. Several job-training programmes were initiated, especially for young people and unemployed women. We should add that these measures were concentrated in regions facing the worst employment problems (i.e. the Northern part of the country and the countryside).

  • Rapid growth in public expenditure: the public sector was turned into a virtual employment-machine, which particularly served to extend job opportunities for women. At the same time Norway witnessed a serious decline in employment opportunities in agriculture and in the industrial sector.

The Project – Two Research Questions

The research project described below was initiated and financed by the Norwegian labour market authorities (Directorate of Work). It had two goals:

  • first, to "map" the participation in two selected job-training programmes for unemployed youth 20 to 24 years of age in 1994, and

  • second, to estimate the employment effects of the two programmes – if any. The two selected job-training programmes covered close to 40% of all job-training for young people 20 to 24 years of age at the time.

The research goals may be translated into a number of specific questions: Who are the participants in these training programmes (i.e. their socio-economic status, educational background etc. compared to the rest of the population of young unemployed? Does their participation improve their employment status? Do women perform as well as men, and how does initial formal education, labour market experience, unemployment experience, etc., influence the prospects of passing the threshold into the labour market for these young people? Does participation to some extent compensate for the lack of formal qualifications and work practice?

The research utilised a quasi-experimental design based on simple random sampling from the two groups of participants, supplemented by a representative sample of registered unemployed young people 20 to 24 years of age who did not participate in any of the two employment measures. Thus the non-participants functioned as the "group of comparison"8. The outcome was measured at the end of a reference period, which was from nine to twelve months after completing job-training (or after unemployment registration) for the majority of the respondents9. We used several definitions of success: ordinary paid work, included ordinary part-time employment, and/or continued education (Eldring and Grøgaard 1996).
The first employment measure – praksisplasstiltaket (the public sector programme) – enables young people to achieve prepared job-training in the public sector. Of the participants close to seven in ten were employed by a public sector employer, and 63% were women. This programme was evaluated in 1993. Try (1993) documented that the participants were significantly better off at the end of the reference period than a comparison group. The two groups of young people were controlled for variation in such individual characteristics (i.e. resources, experiences, and attitudes) and external conditions which might be expected to influence their likelihood of achieving paid work or place in the educational system.
The second employment measure – Noeringslivets fadderordning (the private sector programme – represents an extension of this prepared job-training programme to the private sector. Eight in ten were employed by a private employer, and six in ten were men. This measure was initiated in 1993 by the Confederation of Norwegian Business and Industry (NHO) in close co-operation with the employment authorities, and the major trade unions (LO). This programme was intensively marketed in the business-milieu, and it was presented to the public through a press campaign. The employers association (NHO) undoubtedly put considerable prestige and resources into the implementation and organisation of this new employment programme.
Private and public firms that decided to join the programme got financial support from the Government. In practice this support covered the lion's share of the costs of their participation. In addition the authorities imposed some requirements on the private firms:

  • They had to develop a written individual plan which covered the whole training period10.

  • The firms were supposed to appoint a "godfather" who had special responsibility for the training and the performance of the young adult.

  • The authorities were supposed to follow up – to keep in contact during the process of training and afterwards (i.e. during the reference period).

  • A principle of karens was initiated: if a private firm did not offer the young participant a job at the end of the programme, the employer had to wait for several months, before being allowed to employ a new participant on standard terms, i.e. standard terms of public financial support to the employer and the employee.

  • The authorities decided on the recruitment to the public sector programmes, while the private employers were allowed to influence the recruitment to the private sector training programme.

Research Design

A "quasi-experimental" design (multivariate) is not based on randomisation, but on simple random sampling from a population (universe) of participants and non-participants. Our study compares the outcomes of a specific public intervention, i.e. participation in a prepared public sector and a prepared private sector job-training programme for young people 20 to 24 years of age, and the outcome of non-participation in these two programmes (for unemployed non-participants 20 to 24 years of age). We suspect various types of selection bias:

  • the prosperity to obtain a place on an employment programme may be influenced by (among other things) the social, educational and cognitive resources of the potential participant,

  • the prosperity to accept an offer of participation may be influenced by such resources,

  • the prosperity to respond positively on a postal survey may be influenced by the resources of the respondent.

We found that young people with an above-average score on educational resources were over-represented in both employment programmes, slightly but significantly. Second, the response rate increases successively with an increase in formal education. Third, the response rate among young women was ten percentage points higher than among young men. The average response rate was slightly above 60% (with close to 3,660 respondents in the net sample).

One may to some extent deal with such biased non-responses by applying proper weights. However, we did not do so on the assumption that what we witnessed was "the tip of the iceberg". It is unlikely that we would observe "an icebear below the surface"!11 The empirical "test" does control properly for differences in individual resources such as former labour market experience, formal education, place of residence (county) etc. Still, it does not control for "unobserved heterogeneity".

Empirical Test

In this paper success at the end of the reference period covers the following outcomes: ordinary paid work, included ordinary part-time employment, and/or continued education. The empirical test is carried out in two different ways, which are discussed below.

Figure 2 reveals the effects of the two types of job training as a function of the duration of the training period. Let us start with the group of comparison. According to the figure ("test-line"), we estimate a 50% "natural" outflow from unemployment to ordinary full-time or part-time employment, and/or to further schooling, over nine to 12 months, under the condition that the young people are equipped with average social, cognitive, and educational resources, experiences and attitudes. Then, we observe that both types of participation, i.e. on private sector programmes as well as on public sector programmes are effective, but that that private sector arrangement has by far the strongest positive (net) employment effect.
Figure 3 (comprising Parts A and B) shows that success to a large extent is a function of initial formal education. The figures also reveal that there probably is a statistical interaction between gender of the participant and type of job-training programme, influencing the likelihood of getting either a proper job or schooling during the nine to 12 month reference period. This simply means that the employment effects of the two interventions are somewhat different among men (shown in Part A) and women (shown in Part B). While both the private and the public job-training programmes improve the employment prospects for young women (everything else being equal), the net effect of male participation is only statistically significant among those who have joined the private sector programme. Norway has developed a large public sector since the early 1970s (at present employing about one-third of the labour force), and this sector is clearly dominated by women. According to our computation, the gender difference among the participants on the public job-training programme is statistically insignificant.

Figure 2 Estimated Net Probability of Obtaining Ordinary Work (Part-Time Work is Included) or Further Education During the Reference Period, by the Duration of the Job-Training Period. Logit-Estimate, Liao 1994 (n=3,656)

TEST-LINE: expected proportion of young people in the comparison group (registered unemployed) with the above mentioned hallmarks, who obtain ordinary full-time or part-time work or education during the reference period (close to one year)

SIGNIFICANCE: an indication of the estimated level of uncertainty

PUBLIC: participants in public-sector job training ("praksisplass") with the above mentioned hallmarks or attributes

PRIVATE: participants in private-sector job-training ("Nuringslivets fadderordning") with the above mentioned attributes. The vertical axis shows the estimated probability due to a logit-model, which contains close to 30 variables, included several dummy-variables. The horizontal axis shows the duration of the job-training programme at the beginning of the reference period.

The estimated probability shown in Figure 2 represents an average outcome for men and women, and is based on expected scores on place of residence (county), selected structural features of the local labour market, former work experience, unemployment experience, experience from other job-training programmes, job-search effort (intensity and type of activity), age, social status (parents' educational level), and the educational level of the respondent.

Figure 3 Estimated Net Probability of Obtaining Ordinary Work (Part-Time Included) or Further Education During the Reference Period, as a Function of Job-Training Exposure, Gender and Initial Educational Level of the Respondent. Logit-Estimate
(Part A = Men; Part B = Women)

The horizontal axis shows the initial educational level of the respondent. BASIC: compulsory nine years of schooling

VOCAT1: completed one year of secondary vocational schooling

VOCAT2: at least two years of vocational schooling at secondary level, included completed apprenticeship

GRAMMAR: grammar school (three years)

HIGHER: studies at institutions of higher education, universities included

The labour market for young people seems to be heavily structured according to formal education at present. An unemployed (i.e. non-participating) woman with some higher education had a fairly good chance of acquiring ordinary work after a period of registered unemployment (close to seven in ten). If the young woman completed only compulsory education (nine years of schooling), the chances of getting work seemed to be extremely limited: according to our estimate, only two in ten will obtain ordinary work during a close to one-year reference period if they lack vocational training at secondary level. At this low level of the educational ladder, we do observe somewhat better prospects for men (four in ten obtain a job).
We probably witness two different types of unemployment – among the poorly educated it is fair to speak of a structural of the labour market, among the well educated this is not necessarily the case.


First, we should stress the common denominator of our findings: the formal education of a young job-seeker seems to be an important indicator of success, whether we compare the employment prospects of men and women, of non-participants versus participants, or of people from rural and urban areas.

Boudon's famous analysis from 1974 may throw some light on our observations. He stated that a prominent feature of the educational revolution in modern industrial societies would be a tendency towards more and more education for the same social rewards. The basic reason for this is that it is unlikely that the positional hierarchy of the labour market changes as rapidly as the increase in the educational level of the population. At the same time, this type of "educational inflation" is likely to be part of a Catch 22-mechanism: even though the future prospects of prolonged education decline, the consequences of choosing no education (at secondary level) proves to be hazardous option for the individual. Then the future prospects indeed turn out to be very poor. We do observe unemployment among well-educated young people, but, as of late 1994, the probability of finding solutions to their employment problem seems to be far better than for those who are less well educated.
Second, our findings show that prepared job-training programmes for young people improved their chances of getting ordinary employment. Thus, these measures do make a difference. At the same time, the employment effects of the private-sector programme (Noeringslivets fadderordning) seemed to be more impressive than those of the public-sector programme (praksisplasstiltaket). The figures indicate that there is a significant difference between the outcomes of the two interventions which even survives statistical control for the dominant effects of differences in the educational level of the respondent. Why is this the case?
We believe that these differences in outcomes tell an important story about how these two sectors are structured in Norway at present. Suppose that a private employer is in need of a new employee. For various reasons the employer is not able (or willing) to pay the costs of getting the relevant information about the abilities or skills of a young job-seeker. The employer demands a productive worker, and is not able (willing) to risk employing an unproductive one for a probation period of let us say, six months. When the public authorities decide to subsidise this information through a prepared job-training programme – which as a rule covers the "normal" period of probation in Norway – a majority of the young participants obtain ordinary employment. This was the case for eight in ten participants on the standard programme, which had a duration from six to eight months. Even young people who joined programmes which lasted less than three months improved their chances of getting ordinary work. We believe that these subsidies enable most young participants to demonstrate that they really are productive workers – without any risks taken by the private employer. In the private sector this subsidy seems to create new jobs for young people.
Among public firms and agencies the logic is probably somewhat different: the access to financial and manpower resources are governed by administrative and political decisions, and a rational employer first and foremost sticks to their budget. Of course, a publicly initiated employment-programme poses new prospects of improving and extending the services demanded from this public firm or agency. The employer gets an opportunity to employ subsidised young people, who in most cases do not meet the requirements of formal education, but who are either in need of practice (points) to enter the relevant educational institutions or simply in need of work. No risk is taken, cheap labour is offered, and, in addition, the public employer faces no sanctions (in other words, no obligations) because in the public sector there is no period of karens. This means that it is easy both to get rid of the young person (typically a woman) before the individual becomes too expensive (as an ordinary employee), and to replace the former employee with a new one on the same subsidised terms. In the public sector we probably do observe some displacement effects, i.e. a tendency towards displacement of ordinary employees.
Of course, this way of reasoning does exaggerate the differences between public and private sector, but we do believe that we have pointed at an important mechanism of appointment selection in the public and private sector in Norway.


The evaluation of the "talk programme" warrants the conclusion that the intervention works partly as intended: the number of sickness days is significantly reduced (by 19 days), while the sickness benefit costs are only somewhat reduced. Although the talk programme reduces the number days absent due to sickness, it does not prevent more permanent absence. Nor does it affect the sickness benefits to any significant degree, although the impact is in the expected direction. This latter finding is due to a specific selection process: the intervention interacts with income. The intervention has a better effect on the number of days absent due to sickness among low-income individuals than among high-income individuals.

The evaluation of the job-training programmes for youth shows that these interventions also seem to work, but the private sector programme seems to have much better effects than the public-sector programme. In addition, the effects for men and women are somewhat different.
We have pointed at some possible mechanisms that may explain why the work approach interventions considered here seem to have different effects for different groups of people. Further research is needed in order to establish the empirical status of these hypotheses. The main finding of the evaluations is that the interventions seem to work, indicating that at least some of the measures in Norway's work approach policy are performing as intended.


Boudon, Raymond (1974) Education, Opportunity and Social Inequality, John Wiley & Sons, New York and London.

Eldring, Line and Jens B. Grøgaard (1996) Evaluering av Noeringslivets fadderordning og praksisplasstiltaket (An evaluation of Noeringslivets fadderordning – i.e. the private sector programme – and praksisplasstiltaket – ie. the public sector programme), Fafo, Oslo (in Norwegian).
Esping-Andersen, Gøsta (1990) The Three Worlds of Welfare Capitalism, Polity Press, Cambridge and Princeton University Press.
Grøgaard, Jens B. (1993) Skomaker, Bli Ved Din Lest? En Analyse Av Ulikhet I Utdanning og Arbeid Blant Unge Menn På 80-tallet (Cobbler, Stick to Your Last? An Analysis of Inequality in Education and Work Among Young Men in the 1980s), Fafo, Oslo (in Norwegian).
Hellevik, Ottar (1984) Introduction to Causal Analysis. Exploring Survey Data by Cross-Tabulation, George Allen & Unwin, London.
Judd, C.M. and D.A. Kenny (1981) Estimating the Effects of Social Interventions, Cambridge University Press.
Liao, Tim Futing (1994) Interpreting Probability Models: Logit, Probit, and Other Generalized Linear Models, Sage University Publications, London and New Delhi.
Moen, Kjetil (1991) Ungdomsundersøkelsen (The Youth Inquiry), Arbeidsdirektoratet, Oslo (in Norwegian).
Central Bureau of Statistics (no date) Employment Statistics, Central Bureau of Statistics, Norway (SSB), Oslo/Kongsvinger.
Toby, Jackson (ed.) (1977) Talcott Parsons: The Evolution of Societies, Prentice-Hall, New Jersey.
Try, Sverre (1993) Evaluering Av ArbeidsmarkedStiltak Rettet Mot Unge Ledige (An Evaluation of Employment Measures Directed Towards Unemployed Youth), Institutt for samfunnsforskning, Oslo (in Norwegian).
Gueron, J.M. and E. Pauly (1991) From Welfare to Work, Russell Sage Foundation, New York.
Melsom, H. (1986) "Oppfølging av langtidssykmeldte" NOU, 1986:22, Sosialdepartementet, Oslo (in Norwegian).
Melsom, H. and K. Noreik (1988) "Gjengangersykmeldte og bruk av rådgivende lege ved Skedsmo trygdekontor" Tidsskrift for den Norske Loegeforening, 108:1013-5 (in Norwegian).
Pedersen, J.T. (1993) Arbeid Eller Tvangsarbeid. Resultater av Arbeidsmarkedstiltak for Unge Sosialklienter og Rusmiddelmisbrukere, Universitetsforlaget, Oslo (in Norwegian).
Pocock, S.J. (1983) Clinical Trials: A Practical Approach, John Wiley & Sons, Chichester.
Tellnes, G. et al. (1992) Tiltak for Sykmeldte, Universitetet I Oslo og Rikstrygdeverket, Oslo (in Norwegian).

1 The Salvation Army in Sweden has reported a substantial increase in the number of people benefiting from their charity work, and also that many of these new beneficiaries are not homeless and outcasts, but "ordinary" people.

2 Ironically, one important reason why the level of sick absenteeism has increased may indeed be the work approach itself. If successful, the work approach would result in an increased number of elderly and weak people in the work force, and thus one may expect the level of sick absenteeism to increase (Pedersen 1996).

3 This is an arrangement where people who are sick, but still want and can manage to do some work, are given the opportunity to be at the work place especially suited and easy tasks.

4 The project was financed by the National Insurance Administration.

5 Daily cash benefits for employees equal 100% of pensionable income, and are paid from the first day of the sickness for a period of 52 weeks. The employer covers the costs for the first two weeks. Thereafter the local National Insurance Office takes over.

6 i.e. institutions at university level, or in practice above the level of secondary education and completed apprenticeship.

7 Measured as a proportion of labour force participation due to the ILO-definition. In 1990 the participation rate among teenagers 16 to 19 years of age was around 45%. Most of them combined schooling with part-time employment or seasonal work. The ones staying outside the school system were quite evenly distributed between unemployed (not necessarily as registered) and full-time workers.

8 The ideal research design is based on random selection of entities to a "control group" and a "group of manipulation", i.e. a procedure of selection which is based on a mechanism of chance – of so-called randomisation. In a quasi-experimental (i.e. multivariate) design there is no random selection of members of the two groups of "participants" (those who experience the intervention) and "non-participants" (those who do not experience the intervention). Estimated effects of the intervention are in effect an expression of substitution, of systematic statistical controlled comparison of outcomes between the two groups of research entities. To underline this difference, the "control group" is denoted the "group of comparison", while the "group of manipulation" is denoted the "group of participation" or the "group of intervention" (Hellevik 1984, Judd and Kenny 1981).

9 The range of variation of the reference period for the whole sample was from seven to twelve months.

10 The evaluation showed that this was done in 50% of the cases, but that the young participants as a rule were satisfied with the content of their training, whether it was prepared according to the public demands or not.

11 Of course, this metaphor represents no excuse for neglecting this bias. It simply states that if we observe a correlation between a "battery" of individual resources (e.g. education, former work experience, etc.) and the likelihood of getting work or education during a reference period – which we did – at the same time as the proportion of non-responses in the survey is clearly and consistently influenced by this battery of resources, we will undervalue the relative size of the group of young people facing the most problematic outcomes. Then an assumption about "the tip of an iceberg" (i.e. that problems are worse among the non-respondents) will increase our confidence in a statistically estimated positive effect of this battery of resources on the likelihood of getting work or education. We do, at least, reduce the risk of overvaluing the strength of this association, as results are generalised from the sample to the population of participants and non-participants. If, on the contrary, this "surface" hides a very different "creature" – metaphorically, an "icebear" – it becomes likely that the sample overestimates "the real effects" (which are unknown) of individual resources.

The database is protected by copyright © 2016
send message

    Main page