Flexicurity the Danish labour market model 1

Transcription

1 Flexicurity the Danish labour market model 1 April 2006 Torben M. Andersen Department of Economics School of Economics and Management University of Aarhus CEPR, CESifo and IZA Michael Svarer Department of Economics School of Economics and Management University of Aarhus CAM 1. Introduction The Danish labour market model has been subject to great international attention lately. This is primarily due to the fact that the Danish unemployment rate is relatively low compared to other European countries. It could therefore seem like a natural step for other European countries who have been unsuccessful in bringing down their unemployment figures to try to copy the Danish success story. In their annual report for the Western labour markets Employment Outlook the OECD explicitly encourages especially Germany and France to take a closer look at the Danish model in their efforts to bring down unemployment. The Danish labour market model (the so-called Flexicurity model) consists of three elements: The first element is the flexible dismissal rules (the flex part), second is the possibility of unemployed to receive fairly generous and long lasting unemployment benefits (the security part), and the third element is concerned with the active labour market policy. The flexible dismissal rules ensure that Danish companies are able to adjust their workforce according to changes in the level of production without great costs. The result of the generous social security system is that the individual risk and reduction in income that follow unemployment are reduced. The aim of the active labour market policy is, on the one hand, to provide the 1 We would like to thank Birgitte Højklint for very qualified help in translating this paper. We also thank Niels C. Beier for his comments to a previous version of this paper. The authors bear sole responsibility for any remaining errors. 1

2 unemployed with the qualifications needed to get back in employment, and on the other hand to ensure that unemployed persons are searching actively for employment. Since the beginning of the 1990s unemployment in Denmark has fallen from approximately 12% to approximately 5.5% in This significant drop in unemployment has taken place without either severe wage or price inflation. It is thus natural to investigate whether this favourable development is attributable to the Danish labour market policy. This paper describes the flexicurity model and discusses its strengths and weaknesses in order, among other things, to assess which elements are attributable to the development in the level of unemployment. This paper is organized as follows: In section 2, the development in the flexicurity model over the past 15 years is described, and we discuss whether the development in unemployment is attributable to structural changes in the labour market. Section 3 discusses whether the flexicurity model is capable of maintaining a high level of employment or whether it will result in increased marginalization and by this increased pressure on the public social security system. Finally, section 4 summarises the strengths and weaknesses of the model and discusses to what extent other countries will be able to copy the model successfully. Figure 1: Unemployment rate % Note: Number of unemployed persons as a percentage of the workforce. Source: ADAM-databank, Danmarks Statistik 2

3 2. The Flexicurity Model This section describes the three basic elements of the flexicurity model and provides an overview of the most important changes since Focus will primarily be on the model s capability of ensuring a low level of unemployment. 2.1 Employment Protection Regulations Denmark has very flexible hiring and firing rules. According to the OECD report on employment protection, Denmark is among the quarter of countries with the least restrictive rules on dismissal (OECD 2004). In Denmark, the level of job creation and destruction is relatively high, around 10% (Det Økonomiske Råd (2002)). Furthermore, the level of shortterm unemployment is high. Whether a lower level of employment protection leads to less unemployment is ambiguous both in a theoretical and an empirical sense. There are several factors pointing to the fact that a flexible labour market ensures higher growth since it is easier to adjust the development in the labour market to technological and trade shocks to the economy. Mortensen (2005) stresses that countries with less rigid dismissal rules experienced higher economic growth during the 1990s, and they have thus been able to bring down their unemployment figures. On the other hand, it could be argued that a low level of employment protection reduces the incentive of the individual worker to improve his/her firm-specific skills. As a consequence, productivity, and thus growth, will be lower than in the case of better employment protection. An analysis by Belot et al. (2004) shows that when it comes to maximizing economic growth, the level of employment protection in Denmark is actually too low. Furthermore, the model is better to handle temporary business cycle shocks than structural shocks which may affect the demand for labour in qualitative dimensions. Denmark has a long tradition of flexible firing rules. Since the end of the 1980s, the only change to the Danish employment protection rules has been in relation to temporary agency regulations. Thus, more employment categories have been allowed to use temporary employment contracts. There are no longer limitations on how often the contracts with temporary work agencies can be renewed, and there are no upper limits as to how long one can 3

4 be employed on temporary contracts through a tempororary work agency. Since the use of temporary employment in Denmark is relatively limited only about 8% of total employment it is hard to imagine that these relaxations of the rules have had any significant effect on the decline in the aggregated unemployment. Since the beginning of the 1990s, when the unemployment rate was approx. 12% of the work force, there have been no significant changes to the employment protection area. However, this does not imply that the relaxed dismissal rules have not had an influence on the decline in unemployment. During the same period, a continuous international integration of the product markets, and by this demands for readjustment, have taken place. Whether the Danish labour market has been better than other labour markets at adjusting to the increased international interaction is an open question. 2.2 Unemployment Insurance The Danish unemployment insurance system is characterized by relatively moderate admission demands and eligibility rules for claiming benefits as well as long duration (at present 4 years) (Velfærdskommissionen, 2005). The unemployment benefits system is a contribution financed supplementary insurance in relation to the social security system (cash benefits), but the system is publicly subsidized (marginal costs are publicly financed). The unemployment benefits are determined partly by a replacement rate of 90% of the previous salary and partly by an absolute maximum for unemployment benefits. This combination means that the rate of income compensation is high (90%) for low-income groups, but falling in income (beyond a certain minimum). The average replacement rate is approximately 60%. The structure of the Danish social security system gives rise to a number of questions regarding the economic incentives of unemployed persons to seek jobs. Several analyses have found that parts of the workforce do not gain financially from working as opposed to claiming social benefits (Pedersen and Smith, 2002). However, the constellation of flexible employment protection rules and a generous social security system contributes to the existence of an efficient labour market where the interests of the labour market parties are met. The companies have relatively easy access to adjusting the number of employees in relation to cyclical changes in production and sales, while the trade unions ensure that their members receive compensation if they loose their jobs. Furthermore, an implication of this model is that the 4

5 level of unemployment due to temporary shortage of work is relatively high, which indirectly is an advantage to industries where the need for labour varies greatly over time. In the ongoing debate over employment protection rules where countries with a high degree of employment protection are encouraged to adapt the Danish model it is quite clear that the trade unions in these countries must seek compensation to their members if their job security is reduced as a consequence of relaxed employment protection rules. The effect of this compensation, whether it be in the form of improved possibilities of receiving social benefits or in the form of offers of upgrading and retraining through a more active labour market policy, will be an increased pressure on public spending. At the same time, the increased unemployment benefits will reduce the incentives of unemployed to seek employment or to accept job offers. While the rules on employment protection have been almost the same for a number of years, a number of reforms in relation to the social security system took place during the 1990s. The underlying principle of these changes was that there should be no direct reductions in the level of unemployment benefits. Rather, the incentives to seek and accept jobs should be strengthened. Figure 2 shows the changes in the social security system since the beginning of the 1990s. The eligibility period for unemployment benefits has been reduced drastically. In 1993, unemployed persons were eligible for unemployment benefits for 7 years, and there was a right to re-qualification for eligibility by participation in activation programmes, which basically meant that the eligibility period was unlimited. As a result of the reforms, the eligibility period for unemployment benefits has now been reduced to 4 years, and the right to re-qualification for eligibility against participation in activation programmes has been abolished. At the same time, the possibility of receiving passive social benefits has been reduced. The present labour market policy is based on a so-called right and duty principle, where unemployed are entitled to receive compensation for lost income, but in return they must seek jobs and take part in skills upgrading activities. 5

6 Figure 2 Development in the social security system eligibility and periods of activation As of 1 July 2003 As of 1 January 2001 As of 1 January 2000 As of 1 January 1999 As of 1 January 1998 As of 1 January 1996 As of 1 January Year Passive eligibility period Activation period Note: The right and duty principle in the activation period was introduced in As of 2003 there is no distinction between the two periods. Sources: Finansministeriet (1999), Arbejdsministeriet (2000) and Beskæftigelsesministeriet (2002). Until the latest reform in 2003, the eligibility period for unemployment benefits was divided into two periods - a passive period in which there was no obligation of participating in job activation programmes, and an active period where the unemployed were required to take part in activation programmes for 75% of the time. Since 2003, there has been no distinction between the passive and the active periods. The obligation to accept offers of retraining and upgrading of skills now applies from the first day of unemployment. Unemployed have the right as well as the duty to activation after 12 months of unemployment, and subsequently, the unemployed must receive an offer of activation every six months (this, however, does not apply to recipients of cash benefits with problems besides unemployment). Persons under the age of 30 must receive their first offer of activation after only 6 months of unemployment. Approximately 20% of the recipients of unemployment benefits are taking part in activation programmes, while up to 40% of the recipients of cash benefits who are able and ready to enter the labour market participate in activation programmes. The procedure regarding activation offers varies greatly between municipalities, especially as regards when activation is offered to recipients of cash benefits. 6

7 Despite the fact that unemployed may now be offered activation from the first day of unemployment, the reform in 2003 has not led to an increase in the number of persons activated soon after the first day of unemployment. On the contrary, the elimination of the 75%-rule has resulted in a significant drop in the efforts towards activation of unemployed who are covered by unemployment insurance schemes and who have been out of work for more than a year (Beskæftigelsesministeriet, 2005). The disadvantage of quick activation is the locking-in effect arising from activating unemployed persons, who in any case would quickly find employment. Therefore, the possibility of immediate activation is most often only used in relation to groups of claimants of social assistance. If the eligibility period for unemployment benefits expires before the unemployed person has found regular employment, he/she will be transferred to social assistance. The right and duty principle was implemented here in The compensation for lost income is lower than in the unemployment benefit system and is also dependent on family and financial circumstances, among others. Despite the fact that the level of benefits remains unchanged, the changes that took place during the 1990s reflect a significant tightening of the compensation to unemployed. Figure 3 shows the compensation level for an average worker taking into account the reform changes compared to a situation without reform changes. The figure shows, first of all, a steady decrease in the upper limit of unemployment benefits. This increase can be explained by the fact that the adjustments to the social benefits have not kept up with the development in the wage level. Secondly, the reduction in the eligibility period for unemployment benefits and the fact that the right to re-qualification of eligibility cannot be reclaimed by participation in activation programmes have led to a direct drop in the effective compensation level. 7

8 Figure 3: Compensation rate of unemployment benefits 0,7 0,6 0,5 0,4 0,3 0,2 0, With reform Without reform Note: The figure shows the compensation level for an average worker as the present value of unemployment benefits and cash benefits over a 20-year horizon. I.e. the option value of unemployment benefits and cash benefits in relation to the present value of the average wage. Source: Det Økonomiske Råd (2002). The principle of not reducing the level of benefits has not been followed rigidly. In 1996, special efforts directed at young people were implemented in Denmark. These efforts aimed at young persons below the age of 25 who were eligible for unemployment benefits, who did not have a qualifying education and who had been unemployed for more than 6 months. To this group, the implementation of the reform meant that they had the option of accepting an 18- month education programme with an allowance of 50% of the unemployment benefits, finding employment or receiving reduced social assistance. In 1999, the efforts were expanded to include all young persons under 25 years of age. These efforts had a remarkable effect on the young people s behaviour (see Jensen et al., 2003), and there is no doubt that these efforts can take a great deal of the credit for the, in an international comparison, record low youth unemployment, which by mid-2005 was 3.7% of the workforce between 16 and 24 years of age. On the whole, a significant tightening of the social security system has taken place in Denmark during the 1990s. However, in comparison with the other OECD countries, Denmark is still among the most generous when it comes to unemployment benefits (Nickell et al., 2005). The current Danish government has a declared objective of increasing the incentives to work. In order to achieve this goal, the Government proposed a labour market reform in 2003, which should reduce the level of transfer payments. The reforms were to save approximately 100 million Euros on the public budget. However, as expected, the reform proposal was met with strong opposition from the trade unions. Also the employers were against the proposal because 8

9 reductions in the level of benefits would result in a less flexible labour market. As a consequence, the proposal was withdrawn. 2.3 Activation Denmark has an enhanced active labour market system in which unemployed have access to education, subsidized job training in both public and private companies, job search assistance courses and a number of other measures. The active labour market instrument has been a part of the Danish labour market policy since Initially, the purpose of this policy was to provide a measure which made it possible to regain eligibility for unemployment benefits after the expiry of the original eligibility period. In connection with the introduction of the right and duty principle the activation programmes became the most important measure in relation to obliging unemployed persons to take an active part in improving their chances of employment and to be available for work. The use of activation programmes is quite extensive. Approximately 1.5% of GDP is spent on activation of unemployed every year. It is worth noting that the number of people in activation programmes has not been falling with the decline in unemployment figures. Thus, since 1994, the number of persons participating actively in labour market programmes has been doubled. This means that there has been a significant shift in the prioritization of the Danish labour market policy. From having a social security system with a relatively consistent focus on passive instruments, Denmark is now among the countries in the world with the most extensive use of active labour market programmes. From a welfare perspective, the activation policy is quite interesting as it aims at maintaining the relatively tight eligibility rules and the incentive to work without reducing the rates of income compensation in the unemployment benefit system and the social system. The challenge is how to ensure a flexible labour market and a high employment rate and at the same time reaching central objectives in relation to income distribution policies. There are several potential effects of activation on unemployment. In relation to the length of the unemployment period we distinguish between three effects. The threat effect will typically set in when the unemployed person is approaching the time when he or she is due for activation. The prospect of having to participate actively in an activation programme is not necessarily very attractive, and many unemployed persons therefore choose to leave 9

10 unemployment before they are required to participate in a compulsory programme. The employment seeking activities of unemployed individuals are thus intensified as a result of the risk of having to participate in an activation programme. The locking-in effect refers to the fact that during participation in a programme the job search intensity may be lowered, and the unemployed person may wish to complete the ongoing skill-enhancing activity. The chance of finding employment during the activation period is thus lower than in a system without activation. The post-programme effect refers to the employment situation after participation in a programme. Hopefully, the individual s employability has been increased thanks to the activation programme. There is a risk, however, that the individual as a result of his or her newly acquired competencies chooses to confine his or her search for employment, which could have a negative effect on the individual s chances of finding a job. There have been numerous investigations especially of the locking-in and the post-programme effects of the active labour market policy in Denmark as well as internationally. The conclusion of this literature is that the job creation effects of the system are questionable. In relation to most activation programmes the locking-in effect is quite significant, whereas the post-programme effects in most cases are minor, if not absent. The most positive effects have been registered in connection with unemployed persons in private job training. Other studies have shown that the threat effect of compulsory activation programmes is substantial (see e.g. Black et al., 2003). Rosholm and Svarer (2004) suggest an empirical model that address all three effects in an unified model. They show that the locking-in effect is overestimated and the post-programme effect underestimated if the statistical analysis does not take a possible positive threat effect into account. The argument is that the counterfactual situation is altered when the threat effect is taken into account. If a positive threat effect is not estimated, the hazard rate out of unemployment for individuals who do not participate in activation will be overestimated. This means that when the hazard rate for individuals in activation is compared with this control group the locking-in effect will be overestimated as a lower hazard rate during activation is compared with an overestimated hazard rate for individuals who do not participate in activation. The same line of argumentation leads to an underestimation of the post-programme effect 2. At the same time, their analysis of the Danish labour market shows considerable threat effects of active labour market programmes especially among unemployed 2 See Rosholm and Svarer (2004) for a further discussion of the threat effect and effects in relation hereto. 10

11 men. It is shown that as a result of the active labour market policy the expected unemployment duration for men is reduced by three weeks. The greater part of this effect stems from the threat effect. Thus, as was the case in the above-mentioned analyses, the post-programme effects in this analysis are also relatively modest. However, the analysis points to a potential in the activation policy as the threat effect affects all unemployed individuals. On the other hand, it is still obvious that measured by the large economic costs of maintaining an active labour market policy the positive effects are minor. The analysis points to a political dilemma in relation to drawing up an optimal system. In order to achieve the greatest threat effect the programmes must be as least attractive as possible. On the other hand, the persons participating in these programmes must receive the best help possible to find employment. This seems to call for a measure in the labour market policy which ensures that the unemployed are sufficiently motivated to search for jobs. Possible solutions could be increased focus on ensuring that unemployed persons are actively seeking employment and to apply sanctions against those individuals who do not fulfil the eligibility requirements. Figure 4 shows the development in costs of the labour market policy. As a natural consequence of the general drop in unemployment the total costs have fallen steadily since In return, the costs of the active labour market measures have not been reduced in line with the development in unemployment. The successive changes in the labour market policy during the 1990s (see Figure 2) have resulted in an extensive use of activation measures especially due to the efforts being moved forward in time. This may have had a beneficial effect in relation to increasing unemployed persons incentives to seek employment because of the prospects of having to participate in an activation programme. This effect has especially been observed in relation to groups for whom it was relatively easy to find employment. On the other hand, moving forward the activation poses a risk of placing unemployed persons in a deadlock in the sense that their obligation to participate in activation programmes means that they do not have enough time to search for the right job. In general, however, the extension of the active labour market policy must be credited for having contributed to the drop in the aggregated unemployment. It is clear, though, that it is a very expensive method for bringing down unemployment, especially considering that it is the threat effect that seems to dominate. Thus, a great challenge lies in designing a programme that 11

12 actually increases the chances of employment and which ensures that unemployed persons are allocated to appropriate programmes. Figure 4: Costs of the labour market policy Share of GDP 6 % Active measure Passive measure Note: The passive measure includes costs of unemployment benefits. The active measure includes costs of administration. Source: OECD (2002) and OECD (2005) (OECD has not published any numbers for 2001 and 2002). 2.4 Labour Market Institutions and Unemployment The general conclusion is that changes in the labour market policy have contributed significantly to the drop in unemployment since the beginning of the 1990s. The above shows that tightening has taken place especially with respect to ensuring that the unemployed have economic incentives to seek employment and accept job offers. In relation to job protection, however, there have been no significant changes. The Danish experience conforms to the results of the literature that examines the effect of various labour market institutions effect on aggregated unemployment. The overall conclusion of the literature (see e.g. Nickell et al., 2005 and Belot and van Ours, 2004) is that while the less restrictive dismissal rules do not lead to an unambiguous decline in unemployment, countries with less generous social benefit systems experience lower unemployment. Furthermore, the general tendency is that it is economic incentives that influence the unemployment rate. 12

13 3. Marginalization and the Social Security Net A central issue in relation to the Danish flexicurity model is to ensure that the process of bringing unemployed persons back into employment is sufficiently strong. The flexible hiring and firing rules and the social security system are potentially conducive to readjustment and flexibility in employment. On the other hand, it is crucial that unemployed persons are brought back into employment and do not end up receiving social benefits. The Danish welfare system is characterised by well-developed welfare schemes, which, as a principle rule, are designed to ensure an income for people who are out of work. A system with a well-developed welfare service and a relatively comprehensive social security system can only be maintained if a sufficiently large part of the population is in employment, cf. e.g. Esping-Andersen, The reason for this is that the financing is created mainly through taxation of earned income, and the social security system includes basically everybody without employment. Although Denmark has succeeded in bringing down unemployment figures through the 1990s, a large and increasing part of the population of working age still receive transfer income, cf. Figure 5. It is estimated that approximately half of this group has a reduced capacity for work or health problems. 13

14 Figure 5: Number of people of working age receiving transfer payments, , 1, ,0 1200,0 1000,0 800,0 600,0 400,0 200,0 0, Unemployment Temporary absence from labour market Education Early retirement Health and social problems Note: The figure includes information about educational grants from 1976 and onward. Persons on voluntary early retirement pension falls within Health and social problems. People in activation are placed according to the kind of transfer payment they receive, i.e. either as unemployment benefit claimants ( Unemployment ) or as recipients of cash assistance ( Health and social problems ). Source: Velfærdskommissionen (2004). Figure 5 shows the development of people of working age receiving transfer payments. In the course of one year approximately 1/3 of the people of working age receive some form of transfer payment. The lower layer of the figure shows registered unemployment, while the other layers represent other schemes. Note that the figure does not include pensioners (since 2004, the official pension age has been 65 years, before that it was 67). The figure shows that an increasing number of people of working age receive transfer payments. Although unemployment has been decreasing, the total number of recipients of transfer payments has not fallen. It is worth noting that while the social schemes were extended during the period until 1980, the development has primarily been characterized by an increasing number of different schemes since then. This reflects the fact that people who are marginalized in relation to the labour market receive other types of transfer payments than unemployment benefits. A general characteristic of the development shown in Figure 5 is that labour force participation has dropped from a peak in the early 1980s. In the preceding period, labour force participation grew as a result of increased participation for women. Since then, however, the trend has been 14

15 declining both as regards men and women. A contributory reason for this is the low labour force participation among immigrants and descendants from less developed countries. In relation to ensuring that as many as possible are self-supporting and as few as possible are dependent on social benefits the development has not been all that positive in recent years. Unemployment has been reduced, but the number of people of working age on transfer payments has not fallen significantly. Thus, in terms of avoiding a large burden on the income transfer system, the flexicurity model has not been good enough. 4. Concluding Remarks The flexicurity model is based on three elements, namely flexible dismissal rules, a generous unemployment insurance system and an active labour market policy. This system has basically prevailed both during the high unemployment of the 1970s and 1980s and during the period with less unemployment since the mid-1990s. The most considerable changes to the model during this period were a number of measures that were adopted through the 1990s, which tightened the unemployment insurance system, and which, as a result of the activation policy, tightened the rules regarding eligibility demands and testing. Thus, it is natural to attribute the part of the reduction in unemployment to the tightening of the flexicurity model. At the same time, the development in unemployment is not mathced by a similar success in relation to maintaining a high rate of employment, since, for a number of years, an increasing number of people of working age have been receiving public transfer payments. Many countries have studied the Danish flexicurity model, and the fact that unemployment in Denmark is low compared to most European countries has made it attractive to consider importing the flexicurity model. The Danish experience points towards a number of important factors which must be considered carefully by countries trying to copy the model. First of all, there exists a strong complementarity between this labour market model and the welfare policy as a whole. The model is therefore not directly applicable in other countries with different welfare policies. Secondly, the Danish experience shows that the model does not function solely by virtue of flexible dismissal rules and a generous unemployment benefit system. There is a risk that these two measures alone will generate a situation in which the process from unemployment to employment becomes too weak and thereby undermine the financial 15

16 framework of the model. It is therefore essential that incentives to work rather than being unemployed are present, and an effective eligibility testing must take place. Having all three elements of the flexicurity model to be equally strong is an important prerequisite in relation to ensuring a flexible labour market with low unemployment and high employment. The tightening of the Danish labour market policy through the 1990s took place in the light of favourable macroeconomic trends and falling unemployment, which enabled political argumentation in favour of the need for tighter eligibility demands etc. The question is whether a model based on activation and eligibility testing will prove to be strong enough in relation to greater economic recession and significantly lower employment figures. References: Andersen, T.M., 2005, The Danish Labour Market - From excess to shortage, In Werding, Martin (ed.), Structural Unemployment in Western Europe: Reasons and Remedies, Cambridge MA, London UK: MIT Press. Arbejdsministeriet, 2000, Effekter af aktiveringsindsatsen. Belot, M og J. van Ours, 2004, Does the Recent Success of Some OECD Countries in Lowering their Unemployment Rates Lie in the Clever Design of their Labour Market Reforms?, Oxford Economic Papers, 56, Belot, M, J. Boone & J. van Ours, 2004, Welfare Improving Employment Protection, manuskript, Tilburg University. Beskæftigelsesministeriet, 2002, Flere i arbejde. Beskæftigelsesministeriet, 2005, Analysepapir 5: Den aktive indsats, Service eftersyn - Flere i arbejde. Black, D, J. Smith, M. Berger og B. Noel, 2003, Is the Threat of Reemployment Services More Effective than the Services Themselves? Evidence from Random Assignment in the UI System, American Economic Review, 93(4), Det Økonomiske Råd, 2001, Dansk Økonomi Efteråret 2001 ( Det Økonomiske Råd, 2002, Dansk Økonomi Efteråret 2002 ( Esping-Andersen, G., 1990, The Three Worlds of Welfare Capitalism, Polity Press. Finansministeriet, 1999, Finansredegørelsen 98/99 ( 16

17 Jensen, P, M. Rosholm og M. Svarer, 2003, The Response of Youth Unemployment to Benefits, Incentives, and Sanctions, European Journal of Political Economy, 19, Mortensen, D., 2005, Growth, Unemployment and Labor Market Policy, Journal of European Economic Association, 3, Nickell, S., L. Nunziata og W. Ochel, 2005, Unemployment in the OECD since the 1960s- What Do We Know?, Economic Journal, 115, OECD, 2002, Employment Outlook. OECD, 2004, Employment Outlook. OECD, 2005, Employment Outlook. Pedersen, P.J. og N Smith., 2002, Unemployment Traps: Do Financial Disincentives Matter? European Sociological Review, Vol. 18, No. 3, Rosholm, M. og M. Svarer, 2004, Estimating the Threat Effect of Active Labour Market Programmes, AFN WP Velfærdskommissionen, 2004, Fremtidens velfærd - kommer ikke af sig selv, Analyserapport, København. Velfærdskommissionen, 2005, Fremtidens velfærd sådan gør andre lande, Analyserapport, København. 17