The European flexicurity concept and the Dutch and Danish flexicurity models

In the mid ‐ 2000s, the flexicurity concept was developed into a key EU policy concept. It drew its inspiration from the Danish and Dutch practices to combine labor market flexibility and security. However, the crisis' focus on bring-ing down national deficits and debts left little room to advance the concept. Lately, more emphasis has been placed on the need to take into consideration the social aspect of economic policy ‐ making. Current EU level docu-ments see flexicurity as a guidance for structural reforms. However, the European flexicurity initiatives seem never to have had much impact in Denmark and the Netherlands. There are few accounts of the recent adjustment to the flexicurity models, be it at the EU or at national levels. Therefore, this article assesses the fate of flexicurity by scrutinizing its (adjusted) use as a political concept as well as a socio ‐ economic model. Although the Danish flexicurity model resembles the European flexicurity concept to a large extent, recent reforms have, overall, weakened rather than strengthened the flexicurity model. The Dutch flexicurity model has a narrower focus on normalizing atypical work, while recent reforms support this narrow flexicurity model. Meanwhile, the EU level concept has been changing every year, encompassing a growing number of issues. - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - taken. The EU developments are used to mirror the national policies developments, giving input to ideas of what countries could or should have done. The choice for scrutinizing the fate of flexicurity in Danish and Dutch models is based on a crucial case selection, being most likely to show a balanced flexicurity outcome, even after the crisis. We include the national institutional context when judging how the flexicurity concept was adapted to the national level (Gallie, 2017). The fifth section presents the findings and conclusions.


| INTRODUCTION
The concept of "flexicurity", i.e., the combination of flexibility and security in the labor market, attracted a lot of attention in the year prior to the Great Recession which started in 2008. As such, flexicurity promises the best of two worlds, and emphasizes that flexibility and security do not necessary have to be chosen between-they can be combined (Wilthagen, 1998, pp. 98-202). At the EU level, the culmination was the Council of Ministers' adoption of a set of "common flexicurity principles" in December 2007. However, much of the inspiration came from national levels, especially from Denmark and the Netherlands. The concept itself originates from the Netherlands. One of the aims of Dutch legislation introduced in the mid-1990s was to provide part-time and temporary employees with greater security of employment. In relation to Denmark, the concept has been used to explain a positive dynamic generated by liberal redundancy regulations, high unemployment benefits (UB), and active labor market policies (ALMP).
Since the Great Recession, the attention to the concept seems to have reduced somewhat, both in political circles as well as among academics. At least, this is what is generally perceived (e.g., see Heyes, 2013, pp. 71-86), although it has not been analyzed in any depth to see whether it really is so. Hence, the aim of this article is to analyze the fate of flexicurity, reviewing the changes in its conceptualization as well as its implementation at the EU level and at national levels.
The research question is: "To what extent and how has flexicurity-as a political concept as well as a socio-economic model-been affected by the Great Recession and post-crisis reforms, and has the content of the concept and the socio-economic models changed?".
Methodologically, the article draws upon the authors' previous research, secondary literature, as well as document analyses and statistics. This introduction is followed by the development of the flexicurity concept at the EU level, and two sections that look more closely to the Dutch and Danish cases in order to establish to what extent these models have resisted the crisis and what other changes they have taken. The EU developments are used to mirror the national policies developments, giving input to ideas of what countries could or should have done. The choice for scrutinizing the fate of flexicurity in Danish and Dutch models is based on a crucial case selection, being most likely to show a balanced flexicurity outcome, even after the crisis. We include the national institutional context when judging how the flexicurity concept was adapted to the national level (Gallie, 2017). The fifth section presents the findings and conclusions.

| WHY DID THE EU FLEXICURITY CONCEPT EMERGE AND WHAT IS ITS CONTENT?
2.1 | The process towards formulation of a European flexicurity concept Flexicurity was a part of the European Employment Strategy from the very beginning (Keune, 2008). However, the concept got its highest status when the Commission and some member states started to use the concept as a means to bridge between a political coalition arguing for a more liberal socio-economic regulation and a coalition arguing to sustain or extent the regulation (Mailand, 2010).
The Commission drew inspiration from member states such as Denmark and the Netherlands, where flexicurity could be said to already be reality. In early 2006, the Commission formulated a definition of flexicurity including four "components": "flexible and reliable contractual arrangements, effective active labor market policies, comprehensive lifelong learning strategies and modern social security systems". From the outset, the flexicurity concept-including the balance between flexibility and security-was a matter of controversy. The focus of the Commission in the early phases of the process was especially on moving from job security to employment security, which was in line with one of the Commission's aims, namely to counteract segmentation. However, this focus was also the main reason that several member states and trade unions, especially from the south of Europe, were reluctant to embrace the concept, at least in the early phase of the process (Bekker, 2011;Mailand, 2010).
The Commission's communication, which included the draft common principles, was published in June 2007.
But the draft principles had still to be adopted by the Council in December 2007. In this part of the decision-making process, further changes were made moving the initiative further away from the Commission's original focus, and paying more attention to insiders on the labor market and the social partners (Bekker, 2011). In short, the most important changes compared to the draft principles were: (1) there were no references to the flexicurity pathways in the final version because some member states found them too prescriptive; (2) the references to insiders and outsiders in the labor market (principles one and four) were removed; (3) in principle seven, about the importance of trust and dialogue, the role of social dialogue was emphasized more strongly and the words "socially balanced policies" were added (see Mailand, 2010).
In total, the changes were not path-breaking, but they illustrate that to get everyone on board, the Commission's original path had to be further adjusted. The result was a rather ambiguous concept, flexible enough to allow different actors to read different goals and policies into it (Keune, 2008). See Figure 1.

| The development from the crisis onwards
After the formative years of the European flexicurity concept, two major changes have influenced its further development: the ending of the Lisbon Strategy and subsequent implementation of the Europe 2020 Strategy (2010), and the outbreak of the economic and financial crisis.  (2007) Until 2015, the integrated guidelines of the Europe 2020 Strategy included employment guideline seven on increasing labor market participation and reducing structural unemployment, which, within the total package of the Europe 2020 Strategy, hosts the most explicit reference to flexicurity. It calls to integrate the flexicurity principles into national policies, so as to increase labor market participation, combat segmentation and inactivity, lower gender inequality, and reduce structural unemployment. Although the 2015 revision changed guideline seven into enhancing the functioning of labor markets, it keeps calling member states to take account of flexicurity principles.
While Europe 2020 thus incorporates the flexicurity approach, during the crisis years the concept disappeared to the background. Fiscal and economic issues rapidly gained priority, resulting in stricter EU economic coordination, including stricter depth and deficit rules. This impacted public expenditure severely, also affecting social security, while structural labor market reforms were called for. Especially in the first years of the crisis this left little room for the "security" part of flexicurity, viewing examples at the national level (Heyes, 2013, pp. 71-86;Murphy, 2017, pp. 308-327). At the same time, jobs were disappearing on a large scale.
However, from 2012 onwards a partial strengthening of the EU social dimension started, which might have to do with the alarmingly high (youth) unemployment rates and rising poverty, feeding into recent plans to have a European Pillar of Social Rights. The growing attention to the social dimension has provided the opportunity for the flexicurity concept to re-enter the EU policy debate. Its recurrence began carefully in the Five Presidents' Report on completing the European Monetary Union (EMU) (EC, 2015). This report suggests a common standard for labor markets, which should be developed as part of the completion of the EMU, entailing elements which are quite similar to the "old" flexicurity ingredients (Bekker, 2018). The subsequent Commission Communication on steps towards completing the EMU confirms this idea, while the Annual Growth Surveys of 2016 to 2018 provide more details.
Over time, all four flexicurity elements have been revised. Some have drifted away from the original wording, while others have started resembling the original definition again. Remarkable are the oscillating views both on social security and contractual arrangements (Bekker, 2018). This latter issue was also the most sensitive in the early stages of the formulation of an EU flexicurity concept, a decade ago (see section 2.1). Developments to the concept reflect a continuous search for combining flexible with reliable contracts and developing a new role for social security, being "modern", sustainable, and adequate throughout the life-cycle of citizens.
The re-emergence of the flexicurity concept at the EU level shows that also after the crisis more or less comprehensive ideas on balancing labor market flexibility and security are part of the toolkit. In this respect flexicurity has managed to survive the crisis. The next sections establish how the flexicurity concept managed the crisis in Denmark and the Netherlands, and whether the national developments mirror the EU level developments or not. The Danish combination of high employment rates, low unemployment, low inflation, steady growth rates, low levels of inequality, and well-developed social dialogue attracted attention around the turn of the millennium. A hypothesis for explaining the strong labor market performance was at that time formulated as the Danish flexicurity model. In its most commonly cited version it includes limited job protection, generous UB, and ALMP, which together is said to create a mobile and transaction-friendly labor market (e.g., see Madsen, 2006). 1 See Figure 2.
Flexicurity is also claimed to have been benefitting weaker groups in the Danish labor market. The reasoning includes the expectation that the limited job protection makes employers willing to run the risk of hiring such groups more often than in non-flexicurity countries, because the cost of redundancy would be lower. Hence, the flexicurity model has also been used as an explanation for the comparatively low level of youth unemployment in Denmark.
Since the mid-1990s, youth unemployment caused most concern in the years 2010-12, where it peaked. However, youth unemployment increased less in relative terms than the overall unemployed, and even when it peaked at 14% in 2011, the level was the fifth lowest in the EU. Moreover, the number of young people who are not in employment, education or training (NEETs) is among the lowest in Europe. Nevertheless, NEETs and drop-out rates for youth education do cause concern. Also, Denmark is no exception from the pattern where young people are overrepresented among fixed-term and part-time employees (see Mailand, 2015; Figure 3).
However, from the outset there were a number of basic problems with the flexicurity model as a scientific explanation. The flexicurity model often has been presented as covering all employees on the Danish labor market, although in reality some are much more covered than others. The prototype flexicurity employee is the lower qualified worker in the manufacturing industry and the construction sector, where terms of notice tend to be very short and replacement rates high. Employees in other sectors-as well as non-standard employees-seem to be less covered by one or the other side of the model 2 (see Larsen, Hansen, & Hansen, 2010;Mailand, 2006). Another problem was that although UB-measured by the replacement rates-are, rightly, comparatively generous in Denmark, they are only so for the low-income groups. For the mid-and high income groups, replacement rates are close to the EU average (OECD, 2014). A tendency for employees to sign-up to additional unemployment insurance schemes, administered by the trade unions, has been a way to de facto increase the replacement rates for the mid-and high income groups (Mailand, 2015).
Another problem is linked to ALMP. The public budgets for ALMP are extremely high in Denmark (OECD, 2013), thus supporting its emphasis in the flexicurity model. However, the employment effects of some of the more widespread active measures used, such as job training in the public sector and education, have been seriously questioned in a number of analyses, although the aggregate effect on ALMP seems to be positive (e.g., Det Økonomiske Råd, 2012). Hereby, it is also questioned how well ALMP work as a way back into employment for unemployed people.
Still, the flexicurity model is likely to have contributed to the high mobility and high level of job-creation as well as to the high subjective employment security (Ibsen, 2011;Madsen, 2014).

| References of the concept and links to the European level
The reason that the Danish flexicurity model became a point of reference was maybe not only because of the performance of the Danish labor market, but also because of the model's political appeal as a third way between highly regulated and liberal labor market models. The attention paid to the flexicurity model peaked in the late 2000s. Table 1 illustrates the part of the attention given by the Danish printed media.
Not only the media, but also all main labor market actors-governments, trade unions and employers organizations-signed up to the concept in the mid-2000s, although some did so more enthusiastically than others. Former liberal Prime Minister Anders Fogh Rasmussen liked the concept so much, for at least a period, that he wrongly claimed to have invented it (Mailand, 2006). The trade unions were maybe more enthusiastic than the employers, but also the employers referred to the concept occasionally-and still do, as the Danish Confederation of Employers' EU political goals show (DA, 2017). Nevertheless, all three main actors used the concept more in international presentations than in national debates (e.g., CAOP, Reflect, Hans Böckler Stiftung, FAOS, & Labour Asociados, 2010, p. 89). This might help to explain why a concept that attracted so much attention in international academic and policy circles was not referred to more often in national printed media, even at its peak of attention (Table 1).
Furthermore, although the attention to the general flexicurity concept peaked around the formulation of the European flexicurity concept, the latter has hardly been noticed in Denmark outside academia and narrow ministerial circles.

| The impact of the crisis and policy reforms on the flexicurity model
The economic crisis hit the Danish labor market hard, and unemployment more than doubled, although from a low level around 3% of the labor force. International attention to the Danish flexicurity model weakened somewhat as Source. Denmark: http://informedia.dk/ Note. Web sources and television and radio are excluded. Only national newspapers, regional newspapers and newsletters, organizational and other magazines are included.
Source. Netherlands: http://academic.lexisnexis.nl/ Note. Web sources and television and radio are excluded. Only national newspapers, regional newspapers, magazines (opinions, professional), financial news. Keywords: "flexicurity" and "flexibiliteit en zekerheid". a result (Mailand, 2015). Moreover, the crisis and a number of welfare reforms challenged the flexicurity model.
Below, we present the post-2008 reforms related to the four flexicurity dimensions-i.e., the three dimensions in the Golden Triangle plus lifelong learning.
Contractual arrangements and numerical flexibility in Denmark are in the hands of the social partners and therefore not directly part of policy reforms. Although including collective agreements, not much has changed in relation to contractual agreements in the post-crisis period, but it is worth paying attention to the spread of severance payment, which could be expected to decrease numerical flexibility. Severance payment has traditionally been limited in that only few collective agreements-e.g., in the financial sector-have included it. However, in 2010 severance payment was introduced in the manufacturing industry (Due & Madsen, 2010) and spread rapidly to other private sector collective agreements. This development can be seen partly as a reaction to the crisis, and partly as a reaction to the inter alia, allocates funds for lifting unskilled employees to the skilled level, tailoring the training more to the needs of companies, and increasing the remuneration companies receive when employees takes part in further training (Beskaeftigelsesministeriet, 2017). However, because the reform is so new the effect of it is uncertain, as would be a classification of the reform as being "important".
Regarding ALMP, only minor reforms have taken place post-2008. In the decade prior to the crisis the ALMP system as we know it today was gradually formed. In summary, this is a system where the municipal job-centers have the responsibility for the policy implementation, although within a tight national legal framework. Reforms in the 2000s emphasized work-first policies such as individual counselling, and demands regarding job-seeking activity, but still left comparatively wide scope for "human capital" instruments such as education and training and subsidized work (job-training) (Larsen & Mailand, 2007). During the first years after the crisis hit, some minor youth unemployment and long-term unemployment measures were introduced, and the short-term work arrangement was, for a period, used more than previously. Still, the ALMP policy mainly remained the same, also after the implementation of the latest employment policy reform in 2014, the main aim of which was to simplify the action policy, get rid of "meaningless activation", and support education as an activation measure. Nevertheless, it has to be added that changes have taken place in policy implementation when it comes to the "intensity of activation" (the time the unemployed de facto are taking part in some type of activation activity relative to period registered as unemployed), which was reduced from 29% to 14% between 2011 and 2016 (Danske A-kasser, 2017). Moreover, whereas no major "mainstream" ALMP policy reforms can be pointed to, the reforms of disability pension and flexi-jobs (2012) and sickness benefits (2013) aim at getting people with a reduced capacity to work back into work with a combination of activation measures and "make work pay" measures (see below).
The only one of the four flexicurity dimensions that clearly includes important changes is the social security dimension. Most importantly, an unemployment insurance reform from 2010 reduced the maximum unemployment period from four to two years. Hereby, Denmark was no longer among the member states with the longest maximum benefit periods. Moreover, the reform tightened the eligibility criteria. Although the reform did not change the level of benefit and the replacement rates, income security was obviously reduced.
The reform was introduced by the liberal-conservative government and its allied party in Parliament as well as a party from the opposition, the social-liberals. A tax reform from 2012 added to the anti-flexicurity path by reducing the level of indexation and thereby the replacement rates of the unemployment insurance as well as of other social benefits. The expectation at that time was that the economic crisis would soon be over. According to the government, only 2,000-4,000 persons would exhaust their right to unemployment insurance before they had found a job. However, the government was proved wrong with regard to both estimates, and when a social democratic/social-liberal government came into power, changes were made in the 2012-14 reform. However, these changes were a series of de facto temporary extensions of the maximum benefits periods, not a real change to the reform. The social-liberals had supported the 2010 reform and would not support a new reform including permanent changes.
A UB commission was nevertheless established to formulate a new reform, which was signed in 2015 under a new liberal government. The reform adjusted, rather than changed, the 2010 reform. It included, inter alia, more flexible eligibility criteria by counting in short-term temporary jobs, an opportunity to extend the maximum benefit period by one year (if the persons had taken on sufficient employment during the benefit period), and reduced benefit levels for new graduates without children. The more flexible eligibility criteria represent a small step to include non-standard employees, and could be seen as flexicurity "the Dutch way" (see below). The reform has gradually been phased in and is estimated to reduce the number of persons exhausting the benefit period. Further changes in social security with regard to non-standard employment were made in 2017 with a minor UB reform focusing on hybrid forms of work and self-employment. They harmonize to some extent rules and regulations for employees and self-employed, and between the tax system and the social benefit system, and reduced administrative barriers and burdens for persons combining incomes from employment and self-employment. Still, like the 2015 reform, the changes mildly adjust rather than change the 2010 reform.
In sum, the reform pattern post-crisis shows no major changes with regard to three of the flexicurity dimensions, but important changes regarding the fourth dimension on social security. Although the minor 2015 and 2017 legislative changes show an increased concern for non-standard employees, the dominant "philosophy" behind these and the 2010 reform-and also the above-mentioned reforms of disability pension and flexi-job (2013)

| References to the concept and links to the European level
Like in Denmark, Dutch media attention on the flexicurity concept was the highest in 2007, when EU level discussion led to codifying flexicurity in common principles. However, given the rather low amount of references to it in printed media, one can hardly speak of a "peak" of attention as is the case in Denmark (see Table 1). Instead, the loose term "flexibility and security" has been used much more often in the Dutch debate, and this is related to discussion on the Act Flexibility and Security (Wet Flexibiliteit en Zekerheid) of 1999. This Act is sometimes seen as one of the origins of the flexicurity discussion at the EU level. Allegedly, the term was first used in speeches and interviews by the Dutch sociologist Hans Adriaansens in the mid-1990s in the context of this new Act (see Keune & Jepsen, 2007;Wilthagen, 1998, pp. 98-202). In 1996, a study under his supervision by the WRR (Dutch scientific council for government policy) used the word "flexicurity" as a new combination of flexibility and security (WRR, 1996). Afterwards, the term was especially used in academic circles. Only after a long incubation period did the term enter the European labor market agenda (Auer, 2010, pp. 371-386). Dutch media references to "flexibility and security" were the highest in 2007, when the Act Flexibility and Security was a source of inspiration for EU level policymakers. It was also mentioned in 2012-13, during political discussions on altering the Act in order to get a better balance between flexibility and security (see section 4.4).

| The impact of the crisis on the flexicurity model
The European Commission's claim has been confirmed that the Dutch flexicurity model corresponds to a labor market that has weathered the first years of the crisis. Until 2012, the Dutch labor market seemed relatively healthy, with unemployment rising from 2% in the third quarter of 2008 to 3.8% in the third quarter of 2010 (Eurostat data). Actually, employers welcomed a more relaxed labor market after years of labor shortages. Unfilled vacancies could be filled, and companies continued hoarding labor in expectation of future labor shortages stemming from the aging of society (see CPB, 2011).
Only in 2012 did the unemployment rate start to exceed 5%, after which the Netherlands faced difficulties getting unemployment rates down for a number of years. Unemployment rose to 7.4% in 2014, decreased slowly to 6.9% in 2015, but is now going down rapidly to 4.9% in 2017 (Eurostat data). Although not always an employment "top model", the Commission's evaluation that flexicurity countries managed to preserve employment may be accurate. However, one could argue whether the Netherlands has also been able to preserve a fair distribution between flexibility and security over different groups of workers.
A rapidly growing number of fixed-term employment contracts has not only resulted in more labor market flexibilization, but also started to segment the labor market. This segmentation is perhaps not as severe as in some other EU countries, yet, also in the Netherlands people have been facing longer periods in temporary jobs with decreasing chances to move on to more stable employment. Within the entire workforce the incidence of temporary employment has risen from 17.9% in 2007 to 20.6% in 2016. 3 For young people aged 15-29 these rates are much more dramatic, where a 37% temporary employment in 2007 has increased to 46.4% in 2014 (Figure 3). This means that almost half of the under 30-year-olds have a temporary employment contract. For 44.4% of these employees with temporary contracts, the reason for having temporary employment contracts is that they could not find permanent employment, whereas 41.4% say the contract type relates to a probationary period. Apparently, employers use temporary contracts as a hiring tool. Although the normalization of flex work means that it is not necessarily harmful to be in this type of employment, the philosophy of the Act Flexibility and Security was that employees with temporary contracts would move into steady jobs after a while. However, this stepping-stone effect has become less and less valid viewing the declining transition rates into open-ended jobs. Eurofound (2015) speaks of a Dutch year-to- year transition rate of much less than 20% in 2012, making the Netherlands the second worst performing EU country. Thus, the Dutch model does not necessarily lead to a good performance in terms of a fair distribution of flexibility and security. National legislation is seen as a cause of the high and growing amount of temporary employment and self-employed, as it determines the differences in costs and risks for employers between flexible and open-ended employment relations (see CPB, 2016a).
In spite of the fact that the Dutch aim was to normalize atypical employment, open-ended employment contracts tend to give employees extra benefits, including a higher chance of engaging in training and work-related education.
Furthermore, temporary employment is associated with lower average wages, lower job autonomy, and underutilization of knowledge and capacities of workers (Muffels, 2013). In the long-run, these cumulative disadvantages make temporary employment less attractive to workers. Eurofound (2015) calculates that Dutch temporary workers have one of the highest wage gaps within the EU: more than 20%. For employers, an open-ended employment contract means taking on certain "risky" obligations such as employees being progressively entitled to severance payment in case of redundancy and an obligation to keep paying salary during the first two years of an employee's illness. Only after these two years may the employee become entitled to a public work incapacity payment.
The attractiveness of exploring ways to keep employment outside the traditional open-ended employeremployee relationship has also stimulated the rise of solo self-employment and the invention of payrolling constructions. The latter construction outsources the legal employer-ship, creating a triangular employment relation between the employee, the payrolling company, and the company with the job vacancy. The legal employer is the payrolling company, which thus takes on the risks of illness or dismissal. Still, the payrolling company is little more than an employer "on paper". The employee is selected by, and gets assignments from, the company which offers the job. pay on-call workers for at least three hours per call and questioning payrolling constructions (Dekker, Bekker, & Cremers, 2017, pp. 189-212).
The main changes in the part of employment protection legislation, as implemented by the Act Work and Security, entail the introduction of a transition allowance (transitievergoeding). All workers, also with temporary employment contracts, get this transition allowance if they have been employed by the same employer for at least two years. The allowance may be used for schooling and for moving into another job, and its level depends on tenure.
This allowance is usually lower than the former redundancy payment for open-ended contracts. The expectation is that on average redundancy will be cheaper.
Another change aiming to reduce inequality is that employers can no longer choose via which "route" (Public Employment Service [PES] or court) the worker is made redundant. In court, employees often get severance pay, whereas PES hardly ever decides to grant severance payment. Dismissals on economic grounds are now automatically judged by PES, whereas redundancies on personal grounds appear before court, which should make the system fairer. For workers with a temporary employment contract, insecurity is reduced by shortening the time frame within which open-ended employment needs to be offered. Workers who are employed via payrolling constructions get better protection, having the same level of employment protection as employees who have a direct employment relationship with their employer. Moreover, the long-term use of on-call contracts (zero hours) is limited. In spite of these legislative changes, the Dutch Parliament keeps expressing concerns about the insecurities that Dutch atypical workers face.
Similar to Denmark, there have been no major reforms regarding lifelong learning in the Netherlands. The government introduced a lifelong learning credit in 2017, enabling people up to the age of 55 to borrow money in order to pay for study fees. Also, the tripartite national social partners reaffirmed the relevance of lifelong learning in a recent opinion (SER, 2017). Overall, the Netherlands performs well on lifelong learning compared to other EU countries, although the amount of people engaging in training has been stable for decades, and the access to lifelong learning seems less evident for vulnerable groups such as low-skilled and atypical workers (SCP, 2016).
What is much more interesting is that the Dutch model of normalizing flexible employment, without a deliberate policy link with social security, seemed to be continued. The Act Work and Security was developed in a context of pressures to reduce public expenditure, cutting budgets for re-integration activities, and reducing UB.
Whereas Denmark introduced small reforms to ALMP, the Dutch system did see a major change, which may be summarized as decentralization and budget cuts. Thus, the changes were not so much substantive, but rather concerned changing the governance structure, leaving fewer responsibilities for the national level and giving the municipalities larger roles in providing income and reintegration support. Yet, decentralization was tied to major budget cuts, decreasing drastically the money available to PES and municipalities to offer re-integration services. Eurostat data shows that prior to the crisis, the Netherlands ranked second in terms of ALMP expenditure as a percentage of gross domestic product (GDP), right after Denmark, which was first in the ranking (CPB, 2016b). By 2013, the Netherlands dropped on the list to sixth place. Expenditure on ALMP reduced by 50% between 2002 and 2013, leaving it at 0.8% of GDP. The legislator argued that support at the local level will be more efficient, as municipalities have specific knowledge about regional labor markets and its jobseekers. This was one of the justifications to reduce public spending in this domain. However, compared to other EU countries, there seem to be few funds to guide and coach people towards work or to invest in longer-lasting schooling and training of long-term unemployed (CPB, 2016b).
Regarding social security, the most notable change was the gradual reduction of the length of UB from a maximum of 38 months in 2015 to 24 months in 2019, and the length continues to depend on job tenure. Social partners may agree on giving an additional 12 months of UB in their collective labor agreements. Other changes in social security have been targeted to the most vulnerable groups in society and aimed to get them in a regular job (Borghouts-van de Pas & Freese, 2017).

| CONCLUSIONS
The aim of the article has been to answer the research question: "To what extent and how has flexicurity-as a political concept as well as a socio-economic model-been affected by the Great Recession and post-crisis reforms, and has the content of the concept and the socio-economic models changed?".
We have found that, with inspiration from countries such as Denmark and the Netherlands, a flexicurity concept was developed at the EU level in the second half of the 2000s. This concept included four elements, namely, flexible and reliable contractual arrangements, effective ALMP, comprehensive lifelong learning strategies, and modern social security systems. However, only after concessions were given to skeptical member states and trade unions, could eight ambiguous common principles referring to these four elements be adopted. After the outbreak of the crisis, the Commission succeeded in holding on to the concept as part of Europe 2020, although less attention was paid to it at the expense of the EU's new economic governance. From around 2012-with the socialization of the European Semester-the concept seems to have experienced a partial revival at the EU level. At the same time, the concept has undergone changes.
The Danish flexicurity model-which in its most commonly referred to version includes high income security (generous UB), high numerical flexibility, and ALMP-resembles the EU flexicurity concept to a large extent. The Danish economy and the labor market were hit hard and fast by the crisis. Reforms, especially regarding unemployment insurance, have overall weakened rather than strengthened the flexicurity model. With regard to ALMP, contractual arrangements, and lifelong learning, major reforms have not taken place since the crisis. However, increasing numerical flexibility in a number of other EU countries implies that Denmark has moved from the top to the middle group with regard to OECD-indicators for numerical flexibility. Nevertheless, parts of the flexicurity model seem still to work, as illustrated by continuous high levels of job-creation and job-mobility and low levels of youth unemployment and long-term unemployment. The flexicurity concept is still referred to in the media, but less often than prior to the crisis. The Dutch flexicurity model has a narrower focus on normalizing atypical work. The Dutch economy and labor market were not hit as hard by the crisis as those of Denmark. While Dutch unemployment rose to close to 7% in 2015, it has decreased ever since, and labor shortages are re-emerging. Atypical work has been normalized to some extent, but covers a large part of the labor force. Thus, whereas the Netherlands seems to have resisted the crisis fairly well in terms of employment, labor market segmentation is posing serious challenges to the degree of "fairness".
Contrary to the situation in Denmark, recent reforms support the flexicurity model, especially seeking to find new balances in the first flexicurity dimension of contractual arrangements, and still assuming that people will make the transition into an open-ended employment contract after some time. Other reforms have focused on budget cuts for ALMP combined with more difficult access, or lower entitlements, to social security for certain groups. As in Denmark, lifelong learning has not really been an aspect in Dutch reforms.
Future evaluations of Dutch and Danish labor markets will show to what extent these countries will (need to) incorporate choices for a stronger social dimension into their everyday practices. As yet, the largest reforms in Danish social security and Dutch ALMP have been driven more by austerity and "work-first" demands than by concerns of adjusting systems to the needs of citizens throughout the life cycle, as the EU lately seems to suggest. Will the countries need to start learning from the EU's policy-advices, or will they keep being a model for others to follow?

CONFLICT OF INTEREST
None declared.
ENDNOTES 1 A newer version of the model has changed it into a square by including the voluminous vocational and further education system (e.g., Madsen, 2014).