FIRST-TIME JOBSEEKERS VERSUS UNEMPLOYED WITH PREVIOUS WORK EXPERIENCE: Impact of individual, familial and institutional factors on unemployment durations in Italy

During the last 15 years in Italy long-term unemployment has reached alarming proportions. Italy has one of the highest incidences of long-term unemployment among OECD countries. Since the 1990s, the number of those looking for a job for a year or longer has fluctuated from five to seven unemployed out of 10 (OECD 2001, 2006). The share of first-time jobseekers among the long-term unemployed is remarkable compared to that of the unemployed who have lost a job: 51.6 and 28.2 percent, respectively.¹ Although unemployment in the Italian labour market has been largely studied, little attention has been paid to the effects of the duration of unemployment on (re)employment probabilities, mainly due to the lack of suitable longitudinal data.²

Prolonged periods of unemployment are likely to determine not only loss of income. Being unemployed for a long period of time can also deteriorate the accumulation of work experience and lead to the erosion of work skills. The negative psychological effects of unemployment on the well-being of the individuals have long been recognized (e.g., Clark 1996; for the Italian case, see Figa-Talamanca and Pagano 1988; Preti and Miotto 1999). Recent empirical research on social exclusion among European youth unemployed has found that in countries with high youth unemployment, rigid labour markets but with strong intergenerational solidarity, such as Italy and Spain, most young unemployed are excluded from the labour market but are not otherwise marginalised (e.g., financially, socially, politically) (Hammer 2003). Nonetheless, prolonged unemployment, even when it does not pose a direct and immediate risk of financial deprivation or social marginalisation, is an important determinant of the personal well-being of the unemployed and their long-term prospects in life. Also, it can lead to negative consequences such as delay in new family formation and low fertility rates.

The purpose of this study is to investigate the impact of individual, institutional and family factors on the duration of unemployment spells of first-time jobseekers and unemployed who have lost their job. The paper also aims at assessing the issue of duration dependence – that is, the longer the unemployment spell, the fewer chances of finding a job the unemployed has. From a policy perspective, it is crucial to know whether unemployment leads to further unemployment or whether individuals or groups with specific characteristics are especially prone to the risk of being long-term unemployed. In one case, policy should be directed towards preventing workers becoming long-term unemployed. In the other case, policy should be targeted towards training or retraining measures.

The structure of the article is as follows. In Section 2 the theoretical framework is discussed. In Section 3, the context of unemployment in Italy is illustrated. Section 4 presents the research hypotheses on the determinants of the duration of unemployment. Section 5 describes the data and statistical modelling. Section 6 discusses the empirical results. The final section summarises the findings and draw some policy recommendations.

Over the 1980s and 1990s institutional economists and sociologists have begun to point to the importance of institutional factors in explaining the allocation of the unemployment burden as well as the structure of employment systems. Unemployment is not simply a rate affected by labour market forces. Rather, it is an event concerning the life of individuals who are embedded in social and institutional contexts. These contexts shape and are at the same time shaped by the decisions of every actor involved in the labour market: workers (and their families), firms, trade unions, entrepreneurial associations and the state.

Comparative studies have found that cross-national differences in unemployment dynamics are due to cross-national variations in the way institutional factors interact with economic, or market, factors (for a recent example, see Nickell et al.2005). In particular, this study follows Gallie and Paugam (2000) in arguing that at least three types of factors may affect the experience and duration of unemployment: (1) the welfare system and labour market regulation, (2) the family forms, and (3) the labour market structure. Of particular interest with regard to the first group of factors are the unemployment benefit system and employment protection legislation. For example, generous welfare state provisions to the unemployed are likely to alleviate the negative economic consequences of job loss and to consequently support the jobseekers’ search for adequate jobs. Strict employment protection legislation implies positive or negative consequences depending on the position of individuals within the labour market. It offers high job security for those who have a permanent job but is detrimental for the employment of individuals who have lost a job. Strict regulation of the labour market usually increases employers’ reluctance to hire workers because of the difficulty of dismissing them. In countries where labour market regulation is high such as Italy, jobseekers are therefore likely to experience long-term unemployment spells, which, in turn, are likely to be used by the employers as a signal of deterioration of the workers’ skills. Young unemployed with no work experience are particularly vulnerable in this respect.

The role of the family is equally important for the understanding of unemployment durations. In particular, the degree of stability of the family, extent of intergenerational solidarity and degree of defamilization³ have a relevant impact on the distribution of the burden of unemployment. Clearly, stable family models based on extended intergenerational solidarity and low level of defamilization are expected to protect unemployed individuals from the negative effects of job loss. These models, typically found in Southern-European countries, are also likely (in conjunction with other institutional factors) to increase the risk of unemployment for specific groups of individuals such as younger people and women.

Lastly, the labour market structure and its changes over time are of high relevance, particularly when studying unemployment over a long period of time. For example, in the 20th century the historical transition from rural to industrial economy generated mass unemployment in countries like Italy. More recently, the shift of the occupational structure towards jobs in the service sector and with a high technological content has significantly increased the unemployment risks for individuals with low skills.

To sum up, the central assumption of this study is that the matching process in the labour market depends on the preferences of both supply and demand-side actors but is at the same time constrained by institutional features. These features often determine the distribution of unemployment, namely who has to bear the unemployment burden, rather than the level of unemployment (Esping-Andersen 2000). These factors are also likely to affect the duration of unemployment.

In Italy the labour market is characterized by pronounced regional differences. More than 60 percent of the almost 2 million Italian unemployed live in the Southern regions. The difference between unemployment rates for Northern and Southern regions is striking: the overall unemployment rate for North Italy is 3.5 percent, whereas the proportion of unemployed individuals living in the South is 18 percent of the labour force. In addition to the strong geographical concentration, unemployment is socially structured both in terms of gender and age groups. Female unemployment rate is double that of men (about 12 percent of female labour force compared to 6.7 percent of male labour force). The age gap is even more pronounced. About 27 percent of people aged 15–24 are unemployed, while 7 percent of the labour force aged 25–65 is looking for a job. Long-term unemployment largely reflects the overall distribution of unemployment. Long-term unemployed are predominantly found among young people and women.

With respect to youth unemployment, Italy and the other Southern European countries share many common features: (1) high unemployment rates, (2) prolonged job searches (particularly for the highly qualified), (3) lack of clear vocational qualification signals to employers, (4) lack of developed vocational training systems and little evidence of individual returns to education at labour market entry (Gangl 2003), (5) significant informal economy, and (6) lack of welfare state provision to the young unemployed compensated for by strong family support (Iannelli and Bonmatí 2003). In particular, among the reasons for the high level of youth unemployment are the strict labour market regulation and the low degree of defamilization. The employment protection legislation in Italy is ranked in the middle of the OECD spectrum for the strictness of employment protection (OECD 2004). However, protection against unfair dismissals and collective dismissals is particularly strong and dismissals are consequently difficult and costly. Over time the high regulation of the Italian labour market has increased the divide between protected (employed workers on permanent contracts) and not protected workers (first-time jobseekers and women). Besides, the traditional male breadwinner model family remains the mainstream pattern in the domestic division of labour (Barbagli and Saraceno 1997) and the standard nuclear family with children residing with their parents until adulthood continues to be the most common family model in Italy. Data from the European Quality of Life Survey shows that a remarkable proportion of Italian youngsters aged 18–34 still live with their parents. This proportion is 67 percent for men and 60 percent for women and is by far the highest among the 28 EU countries examined (European Foundation for the Improvement of Living and Working Conditions 2005). With regard to the welfare system for the unemployed, Italy, along with other Southern European countries such as Spain, Greece and Portugal, stands out for low public expenditures on social protection for the unemployed, low levels of active labour market policies and predominance of passive policy measures. Not only is the expenditure on labour market programmes limited but there are also widespread inequalities within the unemployment compensation system. A very small proportion of workers (only those employed in large manufacturing firms) actually enjoy generous unemployment benefits (Cassa Integrazione Guadagni, CIG, and Lista di Mobilità, LM), whereas no benefits are paid to people with no work experience (namely first-time jobseekers) or those who work in small firms.

In addition, Italy is among those countries with weak linkages between individual skills and occupational attainment and earnings, as opposed to countries like Germany where the dual system guarantees far higher labour market integration. As numerous scholars have noted (e.g., Müller and Shavit 1998) educational signalling is crucial in explaining cross-national variations in the distribution of unemployment. Educational systems with emphasis on specific skills and with direct links with employers tend to generate strong skill-based labour markets. Breen (2005) has demonstrated that labour markets with high employment protection and weak educational signalling, such as the Italian one, are likely to exhibit high levels of youth unemployment.

Findings on exits from unemployment indicate that in Italy the first entry into the labour market is a difficult and lengthy process and, unlike in Northern European countries, individual characteristics such as high levels of education or high social class do not seem to reduce the duration of the first job search (Bernardi et al. 2000). It is therefore plausible to hypothesize that such unemployment pattern is strongly correlated with some mechanism of educational signalling. As educational signalling is low in Italy, employers are likely to use work experience rather than education attainment as a signal of productivity. Hence, first-time jobseekers are at a disadvantage, at least until they accumulate work experience. However, educational signalling alone cannot account entirely for the duration of unemployment among first-time jobseekers. School-to-work transitions are embedded in different ‘transition regimes’ depending on the socio-economic structure, institutional settings of education and training, and cultural patterns of the specific country where they take place (Brzinsky-Fay 2007). More factors are therefore likely to be involved. For example, the high protection against dismissal granted to workers employed on permanent contracts and the strong intergenerational solidarity of the Italian family model. As with first-time jobseekers, adult unemployed with previous work experience are subject to the same constraints associated with strict labour market regulation. Yet, unlike the young unemployed, workers who have lost their job can use their work experience as a strong signal of their skills.

Based on the above discussion, the core hypothesis of the study is as follows. I expect that the chances of leaving unemployment for youth unemployed, in particular first-time jobseekers, and those of workers who have lost their previous occupation will be differently affected by individual, family and institutional factors, and will exhibit different patterns of duration dependence. In particular, I expect from first-time jobseekers long spells of unemployment, regardless of their education attainment, social class or age. It is also plausible to expect a situation of negative duration dependence among this group of unemployed – that is, chances of getting a job for first-time jobseekers decrease with time spent unemployed, regardless of their individual characteristics. In other words, first-time jobseekers are prone to the risk of becoming long-term unemployed because of mechanisms associated with time (e.g., scar or discouragement effects) rather than mechanisms associated with education, age or social class.

With regard to adult workers who have lost their job, the hypothesis is that, due to the shift of the occupational structure from low-skill jobs towards high-level skilled occupations in the service sector, they will experience long unemployment spells. However, I expect that their previous work experience will be regarded as a signal of productivity by employers. As a consequence, their re-employment probabilities will not decrease with time spent unemployed but will be strongly influenced by their individual characteristics, in particular their level of education, gender and social class.

This study uses data from the Italian Household Panel survey (Indagine Longitudinale sulle Famiglie Italiane, ILFI). ILFI is a panel study that tracks all of the members of about 4,500 households (over 9,770 individuals), first interviewed in 1997 and every 2 years thereafter. The analysis spans 80 years (1914–1999). It is based both on data from the first and second wave of the panel survey as well as on retrospective unemployment histories collected during the first wave. The sample consists of 8,224 first-time jobseekers (corresponding to 8,224 episodes of unemployment) and 1,690 unemployed individuals who have lost a previous job (corresponding to 2,356 episodes of unemployment).

The effects of individual, familial and institutional factors on (re)employment probabilities are estimated by employing event-history analysis techniques. In particular, I use a piecewise exponential model with period specific effects (Blossfeld and Rohwer 1995) to estimate the hazard rate of leaving unemployment (into employment).⁴ To test the hypothesized different patterns of exit from unemployment, separate regression models for first-time jobseekers and those with previous work experience are estimated. Distinct models for men and women, and models for unemployed in the South of Italy and for those in the Northern regions are also estimated, due to strong geographical concentration and gender differences in Italian unemployment. Besides, to test whether the effects of the explanatory factors vary depending on the length of the unemployment spell, interaction terms between the duration of the spell and the explanatory variables are introduced.

The dependent variable is the hazard rate of transition from unemployment into employment h(t) during a specific month, given that one is unemployed during the month before. Expressed in terms of probability, the hazard rate is the conditional probability of leaving unemployment in the next instant after t, given that the individual is still unemployed at time t.

In this analysis individuals born at very different dates and experiencing unemployment at different points in time are observed over a long time span (1914–1999). This poses the problem of separating age, cohort and period effects. To address this identification issue, age is integrated in the regression models as a time-varying variable, cohort effects are controlled for by introducing dummy variables for each cohort interval selected, and period effects are estimated by using historical macro-indicators (e.g., unemployment rate by year).

Do long-term unemployed workers have low exit rates out of unemployment due to unfavourable personal characteristics or due to scar and/or discouragement effects associated to prolonged unemployment? To test the hypothesis on duration dependence I use results from the baseline hazard function⁵ estimated from the piecewise exponential regression models presented later in this section. Risk functions for the baseline hazard are plotted in Figures 1 and 2. The four risk functions indicate that in Italy the relationship between time spent unemployed and subsequent employment probabilities operates through mechanisms which are different for first-time jobseekers and unemployed with previous work experience.

Figure 1 shows that first-time jobseekers exhibit clear patterns of negative duration dependence: the curves for the baseline hazards monotonically decrease with time and this is the case for both men and women in the South and North of Italy. In other words, the probability of finding a job is high during the first 6 months, but after the sixth month the chances of exiting unemployment decline and remain constant for the following 3 years, after which it declines further.

Figure 1 also indicates that the curves for the baseline hazard of both women in the North and South of Italy are generally lower than those of their male counterparts. Not only do the negative effects of time accumulate over months, but these effects are more disruptive for young women, particularly those residing in the Southern regions. On the other hand, Figure 2 illustrates that the baseline hazards for both male and female unemployed with previous work experience are constant over time, regardless of how long the unemployment spell is. The probability of exiting unemployment is low but does not decline with time. This result suggests that among adult unemployed the probability of getting a new job may depend on individual and familial characteristics rather than on mechanisms (such as scar or discouragement) associated with the length of time spent unemployed.

The results of the piecewise exponential regression models reported in Tables 1–3 help to better disentangle the results from the duration dependence analysis. Tables 1 and 2 show that, among young people searching for their first job, education level (specifically, having a university-level degree) and attending vocational/training courses significantly increase the probability of finding a job only after a lengthy job search. Lower levels of education prove to be more effective in finding a first job, at least in the short-run. The effect of high levels of education on the chances of getting a job becomes positive and pronounced after 12 months for young men and women in the South of Italy, and after 36 months for young men and women living in the Northern regions. In Table 2 the interaction term between tertiary education and unemployment durations shorter than 12 months indicates that, for young women in the South with tertiary education, the hazard rate of finding an occupation within a 12-month job search period is 30 percent lower⁵ compared to the hazard of the reference category (young women with elementary level education). However, interaction terms for tertiary education in Tables 1 and 2 illustrate that in the South, after 12 months, the hazard rates for both young women and young men with a university degree increase by about 60 percent. Counterintuitively, the hazard rate for young women with a university degree, residing in the wealthier Northern regions, rises by 122 percent after a 36-month job search. To explain this striking result, we must bear in mind that in Italy higher levels of education do not significantly reduce the unemployment risks of young workers. In Italy the share of the 25–29 year olds unemployed with tertiary education (and not in education) is the highest among all OECD countries (13.6 percent compared to an average of 4.7 percent) (OECD 2005). In addition, research has shown that during the 1990s, unlike their counterparts living in the South, young people with tertiary education residing in the North exhibit higher unemployment rates compared to those with lower education (Reyneri 1999). This result is therefore consistent with trends previously observed.

The reason for such long unemployment durations among high-qualified first-time jobseekers is not only due to scar or discouragement but may be due to the willingness of school leavers of finding a job corresponding to their education level. Since the mismatch between labour demand and supply does not allow young unemployed to immediately meet their job expectations, they seem willing to wait until those expectations are met. Yet, young people's decisions about labour market entry are not just a mere reflection of labour market conditions but are shaped by a context of socio-economic structures, institutional settings of education and training, and cultural patterns, which vary from country to country and constitute different ‘transition regimes’. The school-to-work transition is therefore also shaped by the features of the model of public welfare. Italy can be regarded as a sub-protective welfare regime (Gallie and Paugam 2000) characterised by the lack of welfare state support to the unemployed with no work experience. Young adults are therefore dependent on their families. Besides, ‘the cultural norm supports an extended dependence residence model, which means that young people are expected to stay with their parents for longer periods of time’ (Benedit et al.2006: 16). Long unemployment spells of first-time jobseekers are thus financially (and emotionally) supported by their families of origin – with whom they live in most cases. Furthermore, some scholars argue that an important aspect of the Italian productive system may be a key factor in explaining the large proportion of highly educated individuals among the youth long-term unemployed: the insufficient demand for high-skill workers (Reyneri 1999). As a matter of fact, the specialization of Italian industries in low-tech sectors and the small size of Italian firms result in a low spending capability for research and development and consequently lead to a limited demand for professional and managerial occupations. However, constraints imposed in Italy by the structure of the welfare regime and peculiarities of the national productive system do not account for the whole picture of unemployment durations among first-time jobseekers. The above explanations leave out a crucial aspect of the Italian career system. Why do Italian young people with a high level of education not make an early labour market entry in positions lower to their expectations and capitalise on professional experience rather than just waiting for suitable jobs? Recent research has confirmed that Italy is characterised by marked rigidity in career mobility. During their life course, most workers are immobile or little mobile (Bison 2002). ‘In Italy, career progression is governed by collective contracts. To date, an egalitarian and corporatist approach has prevailed in the regulation of career steps. As a consequence, work careers display a bureaucratic pattern being linked mainly to seniority, rather than to ability and commitment to work’ (Pisati and Schizzerotto 2002: 4). Also, upward career mobility is found to be higher than downward mobility. This indicates that those who start their career in high-level positions are unlikely to move downwardly within the occupational class hierarchy. For some scholars this may be the reason why educated youngsters along with their families choose to wait before entering the labour market until they find a job that meets their expectations. In this sense, cultural patterns might operate through parents’ aspirations and affect the unemployment duration process. Since entering the labour market in a low-level position often implies to remain in that position or make limited career advancement, parents’ preferences are likely to exert a significant impact on their children's decision to wait for adequate jobs. It is nevertheless difficult to assess what is cause and what is effect.

However, Tables 1 and 2 show that young adults with tertiary education are not the only group experiencing prolonged searches for the first job. Estimates indicate that young people with lower education levels (higher secondary, basic vocational, lower secondary) face substantial risks of experiencing difficult first job searches as well.

Similar results are found in the effect of vocational training. Table 1 indicates that for young women living in the South, attending vocational courses increases the hazard rate of finding a job by 35 percent only after the twelfth month of unemployment. On the other hand, for young women residing in the North, attending vocational courses increases the hazard rate by 110 percent after an even longer job search period (36 months).

Turning our attention to family factors, we find that being married has a negative impact on young women's chances of exiting unemployment. Long-term unemployed married women living in the North have a hazard rate 48 percent smaller than the corresponding hazard for non-married women. This strong negative effect is not found among long-term unemployed women living in the South. This result is likely to be due to the phenomenon, still widespread among young women in the South, of leaving the labour market upon marrying. Not surprisingly, ILFI data reports that a considerable proportion (12 percent) of young women living in the South, who had been searching for their first job for at least 2 years, have actually left the labour market.

Estimates in Tables 1 and 2 also show that the presence of children has no impact on the chances of leaving unemployment. To explain this result, the timing of life events of those who make their first entry into the labour market comes to help. If we look at the median age at which individuals complete their education, and median age at first birth, procreation typically occurs after the first job search for each birth cohort of the sample. Lastly, leaving the family of origin is also found to have no impact on the chances of exiting from either short or long-term unemployment episodes.⁶ This is no surprise as a great majority of Italian youngsters live with their family even when they are employed (OECD 2005). Leaving the family of origin has therefore little or null effect on the chances of getting a job.

In Table 3 results for the piecewise exponential models for adult unemployed broken down by gender are reported. Both regression models indicate that a large number of factors influence the probability of leaving unemployment for those who have had previous work experience. Unlike first-time jobseekers, the hazard rate for unemployed who have lost a job rise monotonically with educational level. This holds true both for men and women even though men seem to have a higher competitive advantage. Estimates show that a male with secondary education face a hazard rate of leaving unemployment 84 percent greater than that of the lowest qualified reference group (male with compulsory education only). Among women with the same set of characteristics, the hazard rate is 38 percent greater than that of their lowest qualified counterpart.

In considering family characteristics, being married represents a disadvantage for women experiencing unemployment spells of any duration. The hazard rate for married women is 23 percent lower than that for single women. Unexpectedly, the number of children has no statistically significant effect on the chances of leaving unemployment. Previous research has found that economic activity rates of both British and German women linearly decrease with an increase in the number of children, whereas economic activity rates of Italian women decline significantly after the birth of the first child (Marauni 1992). This study proves that in Italy being married acts as a deterrent to the transition from unemployment to employment much more than the number of children.When examining re-employment patterns of women, it is crucial to shed light also on the constraints related to intergenerational solidarity. Estimates in Table 3 indicate that the hazard rate of leaving unemployment for women who have received childcare support either by relatives or care services is 36 percent higher compared to that of women who have not received such support. Intergenerational solidarity networks have been long recognised to be crucial to the female labour market participation, particularly to those women who have entered the labour market since the 1970s onwards (Trifiletti 1997). It is a fact that Italian mothers and mothers-in-law have made up for the lack of public child care services for 0–3 year olds (Saraceno 1991).

Contrary to what expected, neither accumulated work experience measured in months nor the number of previous unemployment spells negatively affect re-employment probabilities. Yet, the coefficient for number of unemployment spells in Table 3 indicates that an increase of one in the experienced number of unemployment episodes is associated with a 30 percent for men and 20 percent for women increase in the hazard rate of getting a new job. However, it is worth noting that the proportion of unemployed individuals that have experienced more than one spell of unemployment is limited compared to other EU countries: 18.2 percent have experienced two spells, 5.8 percent have experienced three spells and only 4.6 percent have experienced four spells of unemployment or more. While a number of studies have addressed the existence of potential scar effects resulting from repeated unemployment spells (e.g., Arulampalam et al.2000), in Italy experiencing multiple unemployment spells is not per se a barrier to exiting unemployment. However, it has to be said that the proportion of people in the sample employed with temporary contracts or with no contract linearly increases with the number of unemployment spells they have experienced. Thus, it may be the signal of potential risk of permanent entrapment in precarious work positions.

The effects of unemployment benefits on the probability of re-employment have been widely studied in the Anglo-Saxon literature as well as in Northern European countries. In general, a review of the literature on this topic points out that there is no clear consensus (Devine and Kiefer 1991). However, there is general agreement on the negative effects of prolonged periods of receiving unemployment benefits on exits out of unemployment into employment (Scarpetta 1996). The coefficient for unemployment benefits (CIG or LM) in Table 3 indicates that, within 2 years of the commencement of benefits, men face a hazard rate of exiting unemployment 44 percent lower than that of those who were not in receipt of any benefit. Yet, after 2 years the hazard rate of getting a job is 66 percent higher than for those not in receipt of benefits. This effect is not found among women. Although for women the negative effect of receiving unemployment benefits decreases over time, it remains negative in the long term (even after 2 years).

Long-term unemployment affects 3.3 percent of the EU labour force. Although in most European countries long-term unemployment has declined since its peak during 1997, in the case of Italy, this rate is still above 5 percent (OECD 2004). Besides, the unemployment burden is unequally distributed among the labour force. In Italy, young people aged 15–24 (mostly first-time jobseekers) exhibit the highest unemployment rates as well as the highest long-term unemployment rates. Despite the high incidence of long-term unemployment in Italy, the issue has only been placed on the political agenda in recent years, since the recommendations contained in the European Employment Strategy (ESS), and scant attention has been paid to the study of the effects of the duration of Italian unemployment on (re)employment probabilities. This paper has attempted to address this research gap.

Findings confirm the core hypothesis of this study. In Italy, first-time jobseekers and unemployed who have lost their job are differently affected by individual, familial and institutional factors and exhibit different patterns of duration dependence. Results indicate that even after controlling for gender and strong geographical concentration of Italian unemployment, first-time jobseekers have to face significant barriers to employment. Factors such as high educational level or attending vocational courses exert a positive influence on the probability of finding the first job only in the long-run. Paradoxically, elementary education proves to be more effective in finding a first job, at least in the short-run. Findings also show that unemployed who have lost their job exhibit different patterns of exiting unemployment. Geographical region of residence, educational attainment, and type of job contract of last occupation exert a strong influence on re-employment chances of both men and women with previous work experience. Besides, this study finds strong gender differences in the unemployment patterns. For example, receiving income support benefits does not help women to exit unemployment. Only men seem to benefit from receiving this type of benefits. In fact, factors other than unemployment benefits have a strong positive impact on unemployed women's chances of getting a new job. Among these factors is the impact of intergenerational solidarity. The results demonstrate that child care services and grandparents’ support are crucial in increasing the re-employment probabilities of women who are long-term unemployed. Yet, the ‘other face’ of intergenerational solidarity is that unemployed women who must assist their elderly parents are more likely to experience long unemployment spells.

The analysis of negative duration dependence further supports the core hypothesis of this study and points to an additional difference between first-time jobseekers and unemployed with work experience. While among first-time jobseekers the probability of finding work decreases with the duration of the unemployment spell, the same probability remains low but constant (regardless of the duration of the unemployment spell) among adult unemployed. This means that if first-time jobseekers do not find a job during the first 6 months of their job search, they are more vulnerable to entrapment in long periods of unemployment. On the contrary, Italian adult unemployed are likely to experience prolonged unemployment because of specific characteristics such as gender, low education, lack of vocational training, lack of child care support, assistance to elderly parents.

In terms of policy implications, these findings suggest that in Italy, it becomes crucial to prevent young people from entering long-term unemployment because the risk of entrapment in spells of long duration is high and begins to take effect after only 6 months. As we have seen, a combination of institutional, structural, and family features of the Italian context are likely to contribute to this undesirable outcome: (1) the limited extent to which the education system is capable of sending employers clear signals about the skills of first-time jobseekers, (2) the strict employment protection legislation, (3) the lack of welfare state provision to the young unemployed, (4) the rigid occupational structure and career mobility system, and (5) preferences and aspirations of young unemployed and their parents. In order to increase the chances of entering the labour market for the first time, a possibility would be the use of short-term/temporary contracts. This solution, though, poses the risk of entrapment in employment precariousness. Yet, empirical evidence on the consequences of temporary contracts is mixed and the phenomenon needs further evaluation (Gagliarducci 2005; Zijl and Van Leeuwen 2005). The ‘standard’ positive relationship between education and chances of leaving unemployment, found in studies on North European unemployment, emerges among adult unemployed. Therefore, male and female unemployed with previous work experience and low education would greatly benefit from measures of re-qualification or re-training. Besides, the results indicate that unemployed women would exit long-term unemployment faster if social care services for both children and the elderly were available to them.

Clearly, the study of the determinants of unemployment durations and the issue of duration dependence are just one element of the unemployment experience. It would be desirable that, with the growing availability of panel data, future research could assess whether different mechanisms of leaving long-term unemployment lead, throughout the entire life course, to permanent consequences that differ in terms of subsequent employment career and job tenure, psychological well-being and life transitions.

The study has been realised using the data collected for the ‘Indagine Longitudinale sulle Famiglie Italiane (ILFI)’ project. I would like to thank Antonio Schizzerotto for permission to use the data, and Anthony Cossburn, Christina Mokhtar and Mark Taylor for their comments on previous drafts of this paper. I am very grateful to the reviewers for their valuable comments and constructive suggestions.

Ilaria Covizzi is a Senior Research Associate in the Institut für Soziologie at the University of Basel, Switzerland. She has previously worked as Senior Research Officer at the Social Disadvantage Research Centre and at the Centre for the Analysis of South African Social Policy, in the Department of Social Policy, University of Oxford. Her work has focused on labour market studies (with a specific focus on unemployment), social policy, policy evaluation, and advanced quantitative research methodology.