Flexibilization at the Core to Reduce Labour Market Dualism: Evidence from the Spanish Case

Institutionalist explanations of the high non-standard employment rate in the Spanish labour market have pointed to the relevance of flexibilization at the margin, that is, the deregulation of non-standard jobs. Using Spanish social security micro data, we find evidence that the liberalization reforms of 2010 and 2012 had flexibilization at their core, that is, greater instability due to openended and full-time contracts (erosion of employment protection) and a rise in turnover among periphery workers, increasing the segregation gap. We conclude that there is a high level of deregulation and simultaneous dualization due to the combination of flexibilization at the margin (until 2008) and the core (2010– 2012).


Introduction
There is a heated debate on the trajectories of institutional changes in the employment and industrial relations systems in Western Europe (Howell, 2016). The goal of this article is to analyse the effects of the recent labour market reforms in Spain on the gap between standard jobs (insiders) and atypical jobs (outsiders), that is, labour dualism. More specifically, we (i) explore the evolution of non-standard employment, (ii) research whether there have been any significant changes in standard job stability and (iii) look into any notable transitions from non-standard to standard employment.
Spain is a good case study for several reasons. First, it is considered one of the most segmented labour markets because the atypical employment rate remains among the highest within European countries (Hipp, Bernhardt and Allmendinger 2015), which is a key source of inequality. Second, the institutional reforms that have affected the labour market have been farreaching. This allows us to analyse the complexity of the effects of institutional change. Third, during the post-2010 debt crisis, international authorities pushed for the address of strict internal devaluation policies through labour market reforms and fiscal austerity packages (Baccaro and Howell 2011;Perez and Matsaganis 2018). Finally, these structural reforms are controversial. Some scholars and institutions claim that they have been successful (MEYSS 2013) while others argue that they have failed (Álvarez, Uxó, Febrero, 2019). This is in line with the debate about 'varieties of liberalization' and the processes of liberalization found in advanced or post-industrial societies from a comparative political economy perspective. Thelen (2009Thelen ( , 2014 points out the existence of three types of liberalization trajectories, one of which, dualization, is characteristic of the continental European coordinated economies (Germany and France) and Japan. The other two correspond to the Anglo-Saxon liberal economies (deregulation) and the coordinated economies of northern Europe (embedded flexibilization). In addition, Amable (2003) completed this classification with the introduction of the Mediterranean countries. He indicated that these economies constitute a specific model of capitalism and, as such, they should have a recognized trajectory of institutional change.
However, institutional explanations of non-standard (especially, temporary) employment have habitually pointed to the relevance of partial deregulation, that is flexibilization at the margin while maintaining job protection for core workers, in a classical dualization trajectory (Toharia and Malo 2000). Additionally, it is worth noting that, for some scholars, the structural reforms implemented under pressure in 2010 and 2012 continued on the same path of partial deregulation (Bentolila et al. 2012) and that persistent labour dualism is a result of the remaining over-protection of insiders.
In order to clarify and categorize the institutional change, this article addresses three main research questions. Our first research question is whether there has been a reduction in non-standard employment as predicted by advocates of the reforms. The hypothesis we hold is that there is no tradeoff between the protection of standard jobs and the percentage and number of workers with non-standard contracts, thus questioning the hypothetical success of the reforms.
The second issue that we tackle is whether deregulation has affected core workers. Our hypothesis is that these reforms were a good example of flexibilization at the core, that is, a reduction of standard employment protection legislation (EPL). Particularly, the protection against dismissal (more options and lower cost for dismissal), which constitutes a central tenet of the employment relations system (Emmenegger 2014).
Third, we analyse whether the gap between core and peripheral workers has been reduced following these institutional changes, by studying the transitions of peripheral workers toward standard contracts. We hypothesize that the transition periods are longer and the number of contracts signed is greater (higher turnover), indicating greater hardship when transitioning to a standard contract, in keeping with specialized literature (López-Andreu 2018).
Following a descriptive methodology of working life careers and using Spanish social security statistics and micro data, we are able to detect changes in the stability and security of standard jobs and transitions from nonstandard jobs, by comparing three key years of the business cycle: 2008 (peak), 2013 (trough) and 2017 (peak).
The fundamental theoretical and empirical contribution of our article consists in demonstrating that the austerity measures imposed in Spain during the economic crisis were focused on labour market flexibilization at the core, a typical example of deregulation. However, in contrast to the over-protection of insiders' narrative (Fernández and Martínez 2013), the evidence suggests that deregulation policies have increased instability but failed to reduce labour segmentation. Hence, the strategy of flexibilization at the core has implied an overall increase in precariousness and cast serious doubts about its capability to reduce labour market segmentation (Adams and Deakin 2014). Finally, the unique Spanish trajectory of liberalization is characterized by a twofold path of deregulation and dualization as a result of the various flexibilization policies, which contributes to the extension and qualification of institutional change in mixed-market economies.
The article is arranged in seven sections including this introduction. Section 2 explains the main characteristics of labour dualism and flexibilization policies. Section 3 presents data and methodology. Sections 4, 5 and 6 offer evidence on the evolution of non-standard employment from an aggregate approach, job instability, employment insecurity and segregation of core and peripheral workers, respectively, during the years 2008, 2013 and 2017. Finally, Section 7 summarizes our conclusions.

Labour market reforms: A road to the reduction of non-standard employment?
This section analyses the main characteristics of Spanish labour policy and labour market dualism in recent years, from a labour market segmentation perspective (Doeringer and Piore 1971;Gordon et al. 1981). Accordingly, the labour market is divided into two large segments. On the one hand, the primary market encompasses 'quality' job positions within 'internal labour markets', which are defined as those with higher wages, stability, recognized labour rights, career opportunities, greater union coverage, and so on. Typically, these jobs are for highly qualified or specialized workers in large companies or branches with higher productivity. On the other hand, the secondary market (traditionally identified as precarious jobs) consists of jobs with low wages, labour instability, scarce opportunities for promotion, reduced legal and union protection, lower compensation for dismissal and greater labour risks, among other characteristics. It is well known that positions in the secondary or peripheral segment are occupied mainly by women, young people and migrants, who, as a consequence, find themselves in a more vulnerable work situation and suffering from several gaps, such as lower dismissal costs and wages, weaker social protection and deteriorated working conditions. However, the main features of this kind of job are instability and vulnerability (Kalleberg 2000). According to this framework, the duality between the workers in each segment does not arise exclusively from labour regulation, but is rather reflected in the various forms of hiring. In other words, the characteristics of contracts are the symptom that reflects the structural features of the labour market and hence helps to explain the current labour segmentation.
In the context of a comparative political economy, the liberalization policies in labour market institutions have been summarized in three ideal-typical trajectories relating to three varieties of capitalism (Thelen 2014): deregulation (Anglophone liberal market economies), dualization (Continental coordinated market economies) and embedded flexibilization (Nordic coordinated market economies). However, Spain has generally been considered as a typical example of a Mediterranean mixed-market economy (Amable 2003;Molina and Rhodes 2007). Additionally, it has been considered to have one of the most segmented labour markets even though inequality between insiders and outsiders is lower than in other dualized economies due to the poor condition of the former (Häusermann and Schwander 2012). Nevertheless, the debate about liberalization trajectories, which has focused on EPL, has been increasingly controversial leading to two main interpretations.
On the one hand, some scholars, such as Sola et al (2014), identified two particular stages of the labour policy since the 1980s. The first period (1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994) was characterized by flexibility at the margin, that is the creation of specific temporary contracts, mainly fixed-term, for employment promotion and agency contracts. These legal changes substantially modified the Workers' Charter ('Estatuto de los Trabajadores'). Indeed, until 1984, non-standard contracts could only be used under very specific circumstances, for instance, for seasonal or transitory tasks and in case of need for employers (Garcia-Serrano and Malo 2013). The main objective of these changes was to reduce the extraordinarily high unemployment rate and to improve the employability of certain groups, such as young people, via the introduction of flexibility for labour market entry. This path of liberalization can be understood as 'layering' (Thelen & Streeck, 2005) because the introduction of new types of contracts (fixed-term, training, etc. . . . ) crowds out those employment relationships that are based in open-ended contracts.
However, during the second period (1997-2012) institutional reforms were focused on a reduction of dismissal costs for open-ended contracts and on enhancing the capacity of firms to remove employees or significantly modify working conditions, that is flexibility in the labour market exit. The aim of these reforms was to reduce non-standard employment, specifically the temporary employment rate, as it was one of the highest around the world (Polavieja 2003).
The content of Spanish labour market reforms under austerity is well known (Garcia-Serrano and Malo 2013; Sola et al. 2013, Muñoz de Bustillo andEsteve 2017). In a nutshell, the 2010 reform facilitated the procedures and lowered the costs of dismissal (Cruces et al. 2015). The 2012 reform went much further, significantly reducing job protection legislation (dismissal costs and procedures). Other issues of those labour reforms (Eichhorst et al. 2017), such as unemployment benefit generosity and coverage and active labour market policies, played a minor role that was in line with South European societies. Thus, the recent trend of institutional change is identified as a deregulation process that falls within the varieties of liberalization approach (Picot and Tassinari 2017).
On the other hand, according to the over-protection narrative (Bentolila et al. 2012), workers with standard contracts have very rigid conditions (hindering internal flexibility) and are more protected against dismissal (since their dismissal compensation increases the longer they are employed). As a result, whenever possible, companies lean towards non-standard figures and to turning over workers so that they do not acquire seniority. Thus, outsider workers are the result of over-protected insiders. For these scholars, like Bentolila, Dolado and Jimeno (2012), all labour market reforms occurred at the margin, that is, the deregulation process did not affect the core workers. Consequently, the structural adjustment was mainly suffered by non-standard workers and there was no relevant modification of standard employment.
Based on this over-protection narrative (Fernández and Martinez 2013), the labour market reforms under austerity were aimed at solving structural weaknesses in the Spanish labour market, and reducing wage-setting rigidity and labour dualism, during the critical conjuncture of the economic crisis and one of the highest unemployment rates in Europe. Despite the fact that the 2010 reform was approved by a social-democratic government (Spanish Socialist Party) and the 2012 reform by a right-wing government (Popular Party), both reforms were unilaterally imposed by the Government via a decree law. Consequently, these reforms weakened the social dialogue but they were in line with the previous policy objectives of the largest association of employers (Martín and Alós 2003).
A review of the main changes hints that the reform occurred at the core of EPL. First, the compensation for unfair dismissal was reduced to 33 days of gross salary per year worked with a maximum of 24 monthly payments (previously the limit was 45 days with a maximum of 42 monthly payments). This was important because 'institutions are what actors make of them' (Hauptmeier 2012). The fact that the compensation for unfair dismissal is assessed (quantified beforehand) and that the employer can negotiate it in out-of-court proceedings (such as conciliation, mediation and arbitration) implies that employers could assume this punishment as a sort of cost for exercising their right to dismiss freely or without cause. As such, by reducing the compensation for unfair dismissal, the discretionary power of employers considerably increased.
Second, the 2012 reform created a new sort of contract, the 'support openended contract for entrepreneurs' contracts for companies with less than 50 workers (which includes a large percentage of firms in Spain). Its particularity was that the trial period was set at one year (with no dismissal costs) instead of six months, and tax incentives and social security bonuses were offered in cases where the employee was beforehand unemployed.
Third, similarly, assumptions that dismissals are fair have become increasingly common over the years. Initially, they could be justified only by technical reasons and consent from the labour authority was required. However, the reforms represent a substantial change in the breadth of economic circumstances. After 2012, employers became able to dismiss employees fairly when facing current or expected losses or a decrease in the level of sales or income over three consecutive quarters. Additionally, the administrative procedures for carrying out the layoffs were reduced, as employers became exempt from justifying the company's financial situation. Moreover, dismissals have become the only means of adjustment and the authorization of the Administration for collective dismissals is no longer required.
Fourth, the labour reform has transformed collective bargaining into a weakened tool to limit atypical employment. What is more, the 2012 reform decentralized the wage-setting mechanism at the company level (Malo, 2016), giving preference to company-level agreements over sectoral and provincial agreements (Muñoz de Bustillo and Pinto 2018). The reform greatly facilitated unilateral decisions by employers in collective bargaining (Köhler and Calleja 2017), as well as the application of opt-out clauses by firms. An increased restriction on the criteria used for the extension of collective agreements was legislated and the automatic extension of collective agreements was also substantially reduced.
All in all, our thesis holds that the trajectory of institutional change can be identified as a deregulation process as it constituted a direct assault on traditional EPL institutions, specifically in the form of flexibilization at the core. Notwithstanding, we hold that there is no trade-off between deregulation and decreasing segmentation, because the latter depends on the strategic behaviour of employers (Crouch 2015;Lallement 2011;Polavieja 2003) and over-regulation is not a significant problem in Spain (Prosser 2016). After years of deregulation through layering, employer discretion in hiring and firing has been extended to such an extent that further deregulation will not translate into incentives for using standard contracts. As a consequence, it is possible that a country like Spain shows a high level of deregulation and dualization simultaneously due to the combination of flexibilization at the margin and the core consecutively. The following sections tackle this idea with the help of our three main research questions.

Data and methodology
For analytical purposes, the different forms of contracting can be categorized into two large groups. On the one hand, standard employment (open-ended and full-time) is defined as employees who have an ordinary (throughout the year), full-time (with a working day established, respectively, by labour legislation or collective bargaining agreements) and open-ended or 'indefinite' contract (with no expectation of extinguishment).
On the other hand, non-standard (or atypical) employment covers forms of temporary contracts (including fixed-term, agency, seasonal or discontinuous contracts), part-time contracts (in any of their formulations) and other forms of non-standard contracts (such as training and work experience contracts).
There are indeed two nuances to these definitions. First, not all standard jobs are necessarily held by primary segment workers. Outsourcing strategies and the existence of multi-service companies are additional forms of labour segmentation (Fernández et al. 2016). We will refer to 'core workers' as those who have a standard contract because they have greater employment protection, even if they are not in the primary segment.
Second, we consider only employees, even though there are other forms of non-standard employment including individuals who are solo self-employed (Hipp et al. 2015). The reasons underlying this decision are threefold. First, we focus on the relationship between employers and employees, which indirectly excludes self-employment according to our analysis. Second, self-employees were not directly affected by the labour market reforms approved under austerity. Finally, it is not clear how flexibilization policies affect solo selfemployment. In any case, the definitions of core (insider) and peripheral (outsider) workers are frequently used in the literature without controversy.
Building upon the labour market segmentation framework, we analyse the evolution of non-standard employment using the data of employees from the Spanish General Social Security Scheme (GSSS). The database is therefore restricted to the group of employees who are most potentially linked to standard activities or who are more likely to have standard employment relationships. In short, it is possible to observe the incidence of non-standard employment where it is least likely to arise and to analyse changes in internal labour markets.
Unfortunately, these aggregate data are subject to further limitations as they are not disaggregated by gender, sector, age or company. For that reason, we rely on the micro data from the Continuous Sample of Working Lives ('Muestra Continua de Vidas Laborales', MCVL) database, which is based on administrative records provided by the Spanish Social Security system and made available by the Spanish Ministry of Labour, Migration and Social Security. The MCVL allows us to examine the labour history of each representative worker. The main advantage is that both databases (aggregate data and micro data) are completely compatible. We will focus on the changes in the characteristics of typical and atypical contracts since the year 2008, before the crisis (peak); the year 2013, just after the crisis (trough) and the labour market reforms; and the year 2017, the last one available in this database (peak).
As stated previously, we focus on employment vulnerability because it is the main characteristic of non-standard jobs. Our analysis is based on the following concepts of labour dualism from a working life approach r The instability of core workers is measured as the percentage of core workers laid off before they have worked for three months or one year, respectively, over the total number of core workers dismissed.
r Insecurity is measured as the seniority of the dismissed core workers. r The segregation gap among peripheral workers refers to an increase in the hardship to transit which is mirrored in the following three dimensions: 1. Total transition time is measured as the total number of months that peripheral workers spend looking for employment until they sign a standard contract. 2. Turnover is measured as the average number of non-standard contracts that peripheral workers sign until the achievement of a standard contract. 3. The average duration of non-standard contracts is the ratio of the total transition time over the number of contracts.
Other relevant differences between core and peripheral workers have not been included because they exceed the scope of our framework and would require further research with other databases.

Continuity and change in a segmented labour market
Following the theoretical line of the labour market segmentation theory, the demand for labour is derived from the demand for goods and services. Thus, the labour market is a 'transmission mechanism' in which neither the labour force nor the work actually carried out are homogeneous or individual (Doeringer and Piore 1971;Gordon et al. 1981). In short, job destruction occurs during economic crises and job creation occurs during expansionary phases. How has employment evolved? Table 1 shows the three key years: the beginning of the economic crisis (2008), the end of the recession (2013) and the last year available since the onset of the economic recovery (2017). As shown in the figure, there are two differentiated dynamics.
During the first stage (2008-2013), there was a sharp drop in employment (-18 per cent growth rate with 2.6 million job losses), which started with the onset of the economic crisis. The literature has pointed out (Bentolila et al. 2012) that job destruction is concentrated in non-standard employment while workers with an open-ended contract are less affected by the recession. This implies that peripheral jobs present greater insecurity and instability since they are the ones that are most exposed to the business cycle.
Specifically, one out of every two workers with a fixed-term full-time contract lost their jobs between 2008 and 2013. This is clearly the largest drop in the employment rate in any kind of contract. However, simultaneously  2013-2008 2017-2013 2013-2008 2017-2013 2008 2013 2017 2013-2008 2017-2013 Open-ended Full-time there is an increase in part-time contracts even during the recession (both had a 15.5 per cent growth rate), reinforcing the idea that this type of contract was used to readjust wages in times of crisis. According to Molina and López-Roldán (2015), part-time contracts in the Spanish economy are used as a cost-saving strategy in the context of decreasing aggregate demand. Nevertheless, the over-protection narrative does not take into account that 50 per cent of the total employment reduction occurred in standard employment (a full -9 per cent on a fall of -18 per cent). Despite the even higher contribution of fixed-term full-time, this proves that 1.3 million standard employees were dismissed during the economic crisis. On the contrary, part-time employment had a positive contribution, thus signalling a change in the employment structure.
During the second stage (2013-2017), the economy grows again and job creation takes off. It does so essentially in non-standard contracts, which outgrow standard ones. Indeed, the growth rate of fixed-term full-time jobs was 40 per cent versus the 6.2 per cent growth rate of open-ended full-time jobs. In terms of contribution to total employment growth, most job creation is in fixed-term full-time jobs (6 per cent), followed by fixed-term and openended part-time jobs (3.8 per cent) and open-ended full-time jobs (3.2 per cent). Therefore, the non-standard employment share is higher in 2017 than in the years previous to the economic crisis due to the slower growth of openended full-time jobs.
In short, an initial conclusion holds that high-level non-standard employment persists and has even surpassed the levels existing in the years prior to the economic crisis. This means that the employment created during the economic recovery has been overwhelmingly precarious. Thus, the share of standard jobs has decreased in 3.7 percentage points in the last 10 years, to the lowest rate ever.
Second, non-standard employment, understood as the secondary segment, accounts for 53 per cent of the workers in the GSSS. Provided that atypical situations, such as solo self-employees or special social security systems, have been excluded from the database, the incidence of non-standard employment is extremely high.
Third, Social Security affiliation data show that during the economic crisis fixed-term full-time employment decreased, accompanied by an open-ended full-time employment decline, which has not been adequately considered in previous studies. Consequently, core workers were also affected by the recession and a relevant number of them were expelled from the labour market.
Fourth, the expansion of atypical work is even more pronounced in part-time forms of employment, which reach their maximum rate in 2017 and have increased during the economic recovery. Indeed, the share of part-time jobs has increased by 6.5 pp since 2008, reaching 22 per cent in 2017. Hence, there is a clear shift toward part-time jobs, probably due to a cost-saving strategy in the context of wage devaluation (Uxó et al. 2013). In a nutshell, in aggregate terms, the non-standard job share in 2017 was higher than before the economic crisis. In 2017, there were fewer fixed-term full-time workers, yet this occurred thanks to a significant increase in parttime jobs, with both fixed-term and in open-ended contracts. Consequently, the remaining sections will analyse whether there has been a significant change in labour dualism from a working lives approach.

Transformations in the stability of standard jobs
In this section, we study whether labour instability has also increased in standard (full-time indefinite) contracts, which should theoretically offer higher stability. According to reform advocates (MEYSS 2013), standard employment should have increased and turnover should have decreased. Turning to the former, the share of these contracts is higher than levels recorded prior to the crisis. As for the latter, the average duration of a temporary contract has been reduced, pointing to an increase in instability (Martínez-Pastor and Bernardi 2011; Malo and Cueto 2013).
In the first place, Figure 1 shows that in 2017 the number of core workers who are dismissed (involuntarily expelled in an individual or collective dismissal) before reaching one year of seniority is approximately one out of three. This value has grown since 2008 (28.8 per cent) and 2013 (24 per cent), implying not only that the number of typical contracts signed per year has fallen, but that there has also been an increase in short duration open-ended full-time contracts.
In addition, we observe a change in the 'time until dismissal' distribution between 2008 and 2017. First, the number of employees being fired within the first three months reached 53.3 per cent of all dismissed workers in 2017; an increase of 6 pp since 2008. Taking into account that the trial period according to the Worker's Charter is set to six months for technicians and two months for the remaining categories (trial periods may be modified under a Collective Agreement), many core workers were dismissed without compensation. Unfortunately, the MCVL lacks information on severance payments.
In the second place, the percentage of dismissed core workers with greater seniority increased between 2008 and 2017. Despite an increase in concentration in both tails of the distribution, the average seniority of dismissed workers has increased. In this sense, insecurity has expanded to standard jobs, since employers are more inclined to expel this kind of worker, instead of looking for alternative solutions to dismissal as in the period previous to the labour market reforms. This might be a consequence of the reduction of exit barriers, which is mirrored in the decrease in the percentage of typical employment since 2008.
These results are coherent with the findings of López-Andreu (2017), who holds that legal changes have led to the erosion of the protection associated with the standard employment relationship and provoking unexpected turning points and career breaks as well. Eurofound (2019) carries out a comparative study of Germany, France, the United Kingdom and Spain. According to their findings, Spain is characterized by the group of workers with the lowest standard trajectory. Its labour market shows limited upward mobility and increasing difficulty in achieving a standard career for large numbers of people who are in transition between unemployment and nonstandard forms of employment, with relatively poor working conditions (typically temporary contracts).
All in all, the evidence undoubtedly points to an increase in both of the proposed dimensions, instability and insecurity, among the most stable jobs after the 2012 labour reform. However, in order to identify the causes of this dynamic, one requires a more detailed approach. Table 2 outlines the evolution of these two dimensions, with a higher level of disaggregation.
First, disaggregating by gender, the percentage of core workers dismissed before they have worked for three months or one year increases for both women and men, but especially for the latter (35.8 per cent are fired before one year and 15.5 per cent before three months). On the other hand, the seniority of women with standard contracts who were fired has raised significantly (35.6 per cent). Thereby in 2017, there is no relevant gender seniority (both stand at 49 months).
Second, we break down the data by the size of the companies: small and medium-sized enterprises (SME) and companies with more than fifty workers. The latter have sharply reduced the percentage of fired core workers, while the former have increased this percentage dramatically (9 and 7.1 pp, respectively). Hence, 37 per cent of core workers from SME are dismissed before one year and 16.4 per cent before three months. However, it is in the largest companies where insecurity is currently highest since the average time until dismissal has boosted substantially (36.7 per cent), meaning that the number of workers with high seniority being fired increased, provided that their employment protection deteriorated. Third, if we consider differences among age groups, young people (25-35 years old) are dismissed more speedily than other group. In 2017, 40 per cent of them are fired before one year and 16.6 per cent before three months (an increase of 9.3 and 6.9 pp since 2008, respectively). The average number of months until dismissal has also risen for all core workers between 35 and 65 years old, especially during the economic crisis.
Fourth, in an effort to ensure that there is no sectoral bias, we group the sectors into six broad categories: business services, construction, hightech manufacturing, low-tech manufacturing, personal services and primary sectors and utilities (see Table A1).
Similar increases in instability are observed in high-tech manufacturing, low-tech manufacturing and personal services. It is worth noting that in construction almost 50 per cent of core workers were dismissed within a year and 21.6 per cent in the first three months (an increase of 17.5 and 12 pp, respectively). Moreover, instability has grown in high-tech manufacturing (60 per cent in 2013 and 40 per cent in 2017) and in construction during the economic recession (90 per cent between 2008 and 2013).
Finally, one possible explanation for these changes is that the extension of atypical employment has been the consequence of skill-biased technical change derived from digitalization, automation, or even other processes such as globalization or tertiarization, which can foster a job polarization dynamic. Table 2 shows that the increase in instability has taken place in all sectors, except for the primary sector and utilities. Following the classification of Molina and López-Roldán (2015) into different degrees of qualification and skill, not only did the number of low-skilled jobs increase but so did their instability beginning in 2008. We find that low-skilled workers are those who suffer from greater instability, 42 per cent are dismissed within a year and 20 per cent in the first three months. However, highly skilled workers are those who underwent a greater increase in their insecurity in 2013 (40.9 per cent) and in 2017 (34.4 per cent) given that they were fired despite having greater seniority.
In a nutshell, there is a general increase in instability and insecurity at every disaggregation level. The main difference is that instability has increased to a larger extent among young people, SMEs, the construction sector and low-skilled workers, which is probably related to the Spanish growth model (Buendía 2018). Additionally, insecurity has increased more markedly in large companies, high-tech manufacturing and highly skilled workers. Therefore, core workers with these characteristics are currently more exposed to dismissals than they were previously despite their greater seniority, indicating lower exit-barriers in the Spanish labour market.

Labour market trajectories: Increasing hardship in securing a standard job
Contrary to the claims of proponents of the labour market reforms, deregulation in the Spanish labour market has failed to reduce dualism. As seen in the previous sections, the increase in instability in open-ended contracts mirrors one aspect of the negative effects of the labour market reforms. The other negative effect mentioned is that of the increase in insecurity, as the backlash in EPL has fostered the dismissal of core workers with accumulated seniority.
There is yet another negative effect of this deregulation strategy that must be put forward. Spanish labour market reforms have extended the transition periods from non-standard contracts to standard contracts as well as increasing the number of contracts signed prior to signing a standard contract (turnover). In other words, transitions from non-standard jobs to standard jobs have become more time-consuming and are characterized by a higher degree of turnovers, widening the gap between peripheral and core workers.
The changes in the segregation gap between core and peripheral workers can be observed in three dimensions. In this section, we deal with the evolution in the three key years of both the transition time peripheral workers spend until they sign a standard contract and the number of non-standard contracts they sign until the achievement of a standard contract.
A third dimension must be analysed as it emerges from the interaction of the two aforementioned dimensions. Depending on the evolution of the transition time and turnover, the average duration of non-standard contracts varies, thus affecting the segregation gap. The 2012 labour market reform increased the dismissal compensations of temporary contracts from 8 to 12 days, which may have led to an increase in the stability of this kind of contract. Table 3 offers evidence on these three dimensions, disaggregated into various categories. In aggregate terms, the time peripheral workers take to transition from a non-standard job to a standard one increased by 35 months from 2008 to 2017, reaching a total of 97 months. Turnovers also augmented substantially from nine non-standard contracts during the transition period (2008) to almost 15 contracts (2017).
One must note that the increase in turnover is concentrated in the recent expansionary period (2014)(2015)(2016)(2017), whereas the rise in transition months is more equally distributed between the recession (2008-2013) and the expansion. As a result, the average duration of non-standard jobs increases in 2013 with respect to 2008 because of the freeze in recruitment. On average, non-standard contracts lasted almost seven months in 2008, as compared to 2013, when they lasted almost three quarters. Once economic growth turned positive, turnover drastically increased to more than 14 contracts in the transition period and the average duration of non-standard contracts fell back to slightly below 2008 levels.
Breaking down the data into different categories such as gender, company size, sector, age group and skill level yields the following findings. First, the Source: Own elaboration using the MCVL segregation gap evolution differs by gender, since, for men, the increase in total months until transition is greater than that of women (36 months vs 34 months throughout the period) and even surpasses the latter figure in 2013. As for turnover, on average, women and men signed the same number of nonstandard contracts in 2008 (a total of nine contracts). During the recessionary period, turnover for men increased by one contract while the number of nonstandard contracts signed by women remained at 2008 levels. During the period from 2014 through 2017, the gender turnover gap augmented to two contracts, yet the increase in turnover affected both genders reaching a total of 15 contracts for men and 13 for women. The evolution of labour trajectories might also diverge depending on the size of the firm. Larger firms that benefit from more consolidated positions may opt for standard contracts. As shown in Table 3, the increase in months to transition and turnover is a common feature across all companies. Moreover, workers in smaller firms underwent a higher increase in the total months to transition throughout the period than workers in larger firms (reaching 100 months for the former and 91 for the latter in 2017). As for turnover, labour market reforms generated a turnover gap, since the number of contracts signed by a small firm worker increased by one in 2013, whereas the average worker in a larger firm signed the same number of contracts as in the peak year (nine contracts). Regarding the period 2014-2017, there was a generalized increase in turnover of five contracts.
Turning to a sectorial breakdown, labour trajectories of workers in the primary sector and utilities were those most affected by the reforms. In terms of the transition duration, the transition time increased by 94 months for a total of 153 months in 2017. In terms of the level of turnover, a similar path can be observed. The number of contracts increased from nine in 2008 to 23 in 2017. Contrary to the remaining sectors where turnover rose only during expansion, the number of contracts signed by these workers was already on the rise during the recession, reaching 15 contracts in 2013.
As can be inferred from the disaggregation by groups of age, younger workers suffered the highest increase in turnover (from 10 contracts to 16). Meanwhile, the highest rise in the transition time was experienced by workers 45-55 years old, followed by workers aged 35-45 and those aged 55-65, reaching a total of 114, 119 and 113 months of transition, respectively.
Finally, a clear difference can be observed when breaking down the data by skill level. Even though there was a generalized increase in transition difficulty for all workers, those who were highly skilled were somewhat more protected, as the total months to transition in 2017 remained below the total average (79 months vs 97 months). Moreover, the turnover of highly skilled workers only increased by two contracts in 2017. The labour trajectories of low-skilled workers were the harshest in 2008 and turned even harsher by 2017, with a rise in the transition duration to 106 months and an augmentation in turnover to 19 non-standard contracts. The evolution of the difficulty of skilled workers to transition was in line with the total average. To sum up, the increase in difficulty when transitioning to a standard contract is a generalized feature of all peripheral workers, regardless of the different categories analysed in this section. All in all, the process of deregulation in the Spanish labour market led to a widening of the segregation gap between core and peripheral workers, which is mirrored not only in the increase in instability and insecurity of core workers but also in the changes in labour market trajectories of non-standard workers.

Conclusions
First, in structural terms, we quantify non-standard employment at about half of total employment, whereas in dynamic terms there is a temporary fall in fixed-term full-time jobs. This is caused essentially by the economic crisis (workers with this type of employment relationship being the first to be expelled from the labour market) and the number of part-time jobs has gone up dramatically since 2008.
The evidence retrieved from the trajectories of core workers shows that a significant percentage of them, with theoretically stable contracts, were dismissed after a period of only a few months and, additionally, the share of employees being fired within the first three months in 2017 exceeded 2008 figures. This suggests that the instability of periphery workers has spread to core workers. Additionally, core workers are being expelled with larger seniority than they were previously, thus increasing their level of insecurity (López-Andreu 2018). It also reinforces the idea that labour market reforms, through the cut in severance payments and the loosening of fair dismissal conditions, facilitate the layoffs of workers that had previously accumulated a greater degree of employment protection. To our knowledge, this is the first article that finds evidence about this relevant issue.
Despite the fact that the inequality between insiders and outsiders in Southern European countries is lower due to the poor condition of insiders (Häusermann and Schwander 2012), in Spain, the gap between the primary and secondary segments is increasing. Hence, for a peripheral worker, it is ever harder to transition toward a standard job, while it is ever easier to be expelled once he/she signs a standard contract. Accordingly, there is no significant change in the composition of the labour structure and, thus, the segregation gap between the two groups of workers persists.
The labour policy of deregulation at the core is therefore not the driver of the remarkable improvement in labour market performance but rather the trigger of an increase in labour dualism and instability. Thus, the trend is toward an 'all precarious' situation (López-Andreu 2017) where the trajectories of core workers become more insecure. This shows that turnover practices, which are deeply rooted in fixed-term jobs (Polavieja 2003) and even in the most stable contracts, have not been significantly reduced by the labour market reforms.
Especially, given that after decades of flexibilization at the margin, temporary and part-time contracts became a common practice among employers with the aim of reducing labour costs (Fernández et al. 2016;Molina and López-Roldán 2015). Such practices were enhanced by the wagedevaluation policy in place since 2010 (Álvarez et al. 2019). Additionally, labour relations literature has revealed that the capability of collective bargaining to reduce labour segmentation is decreasing as a consequence of decentralization policy (Muñoz de Bustillo and Pinto 2018). Hence, there is a strong path dependence in the segmentation practices of employers because there is no countervailing power to reduce labour dualism.
Our contribution is therefore twofold. First, we show that an increase in core flexibility is not associated with a reduction in non-standard employment and as a consequence, there is no trade-off between them. Second, there is evidence that a simultaneous process of deregulation (particularly, in the core) and increasing labour segmentation is plausible. This is a relevant finding for the 'varieties of liberalization' debate in mixed market economies, because the Spanish case shows that, after decades of flexibility at the margin (dualization), a labour policy of flexibility at the core (deregulation) leads to a labour market that is simultaneously dualized and precarious. Hence, the type of liberalization trajectory we have analysed is deregulation with dualization. Moreover, this raises serious doubts about the capability of deregulation to reduce labour dualism. The results of 30 years of deregulation in Spain are notably disappointing.
As a corollary, the strategic behaviour of social agents, mainly employers, must be taken into account (Pulignano and Doerflinger 2018). Given that the competitive strategy was based on a reduction in labour costs, flexibility policy has achieved that objective (Cruces et al. 2015), while failing to reduce labour dualism. As such, we can conclude that the complementarities between the strategic behaviour of employers, the growth model and the path dependence of labour policy are key factors for explaining the underachievement of previous reforms.
Consequently, it is necessary to consider alternative ways to reduce the segmentation level of the labour market in Spain. The increase in the discretion of employers when hiring and firing has reduced the costs and flexibility conditions gap between standard and non-standard contracts (due to the later labour reforms). However, further research is required to analyse the effects of employer strategies on labour market segmentation from a comparative perspective.
Final version accepted on 30 March 2020