Analysis of Long-Term Non-Employment in Italy, Spain, and Germany

Abstract

This chapter proposes a new approach to the analysis of long-term non-employment, its duration and main causes using administrative longitudinal databases. Non-employment denotes able people who have never worked or have withdrawn from work for whatever reason and who are not actively looking for work. In this paper, we estimate the number of long-term non-employed (LTNE) in Italy, Germany and Spain LTNE duration and some of the LTNE’s main characteristics. Non-employment duration averages 10–15 years for people in their 30s and may reach 20–25 years for the 45–50-year-old.

1. Introduction

The long-term non-employed (LTNE) are people in working age who have had spells of regular employment at the beginning of their careers, stopped working for whatever reason and became unemployed but no longer succeed to regain a job in the official economy for a very long time. For many, non-employment turns out to be a lifetime condition. LTNEs differ from the officially recognised “out of the labour force” category which includes young people at school as well as old workers after retirement.

Non-employment duration averages 10–15 years for people in their 30s and may reach 20–25 years for the 40–50-year-old. LTNE has drawn little attention in the academic literature, although its long run consequences imply dramatic changes in individual lifestyles, family and childbearing projects, levels of poverty and welfare at large.

This study proposes a new approach to the analysis of male long-term non-employment, its duration and main causes using administrative longitudinal databases of all workers, including the self-employed. The observation period starts in the late 80’s and ends in 2012. The key indicator is “labor market survival”: it denotes the time period that elapses since one’s first hire or initial job spell as self-employed until his definite exit from work. If he leaves a job and joins a new one after one or more spells of unemployment, he will be counted as a survivor. LTNEs are those who do not survive in the observation period 1979–2012. The length of their non-employment can be observed from the administrative longitudinal data.

The chapter is organised as follows: Section 2 illustrates the main aspects of labour market developments in our countries. Section 3 provides a short survey of the relevant literature. Section 4 illustrates the administrative databases. In Section 5, we present estimates of long-term non-employment magnitude. Empirical survival schedules are displayed and discussed in Section 6. Section 7 presents the estimates of long-term non-employment duration. Section 8 addresses the question of the end destination of the non-survivors. Section 9 is dedicated to an exploration of ECHP (European Community Household Data) that provides additional information on non-survivors. Policy implications and conclusions close in sections 10 and 11.

2. Labour market developments

Unemployment figures have reached extremely high levels all over Europe: between 2008 and 2014, the number of unemployed individuals increased from 17 million to 24.8 million (EU 28). In addition, Eurostat indicates the existence of over 16 million ‘inactive but willing to work’, in addition to the participation rates which include the latter whether or not willing to work. Not only is the magnitude of unemployment, official or unreported, a source of concern, but also – and perhaps even more – is the problem of long-term non-employment and its duration among able people of working age. In this study, we find that non-employment duration averages 10–15 years for men in their 30s and 15–25 years for the 40–50-year-old. Our explorations are limited to the male component of the workforce. The necessary data for the study of women workers were incomplete at the time of the inception of this study.

The numbers involved should be a source of major concern. It is surprising that the literature has so far paid modest attention to the dramatic duration of non-employment among people of working age and its far-reaching implications for social policy. Long-term unemployment has been the subject of innumerable academic studies referring to the official (Eurostat) data that define long-term as 1 year+ (seldom 2 years+). We claim that the length of non-employment duration found in this exploration poses more serious and qualitatively different causes for concern.

The process leading to increasing LTNE was underway in various EU countries before the dramatic downturn of 2008 [1]. We compare long-term non-employment (LTNE) in three countries with different labour market institutions: Italy, Spain and Germany in 2012. These countries provide an interesting example of how institutional arrangements impact the magnitude of unemployment and inactivity. Unemployment insurance in Spain and Germany is available for almost all the unemployed (Germany’s program being more generous than Spain’s), leading many dismissed workers to promptly register as unemployed, not counted as LTNE. In Italy, instead, adequate unemployment benefits are offered only to employees of large firms (C.I.G. Cassa Integrazione Guadagni) and do not cover other workers who may become LTNEs. Therefore, the magnitude of LTNE is different, Italy’s being higher than Germany’s and Spain’s.

Part of the responsibility for these developments resides in the reforms advocated by the EU Commission and implemented to different degrees in many member states since the 1980s. While aimed at enhancing youth employment opportunities by lowering entry wages and increasing contract flexibility, they often provided employers with incentives to pursue strategies of rapid turnover and continuous replacement of young people that impeded a sound development of human capital.

The EU unemployment situation and the increasing precariousness of work and jobs at the beginning of the new millennium are well documented and require few comments. In 2014, the EU-LFS reported estimates of the “inactive, but willing to work” at 16.1 million in the EU (28). Italy’s rate was more than double the EU average and far above all the larger EU countries, including Spain whose unemployment rate was much higher than Italy’s.

Many of the inactives are presumably discouraged unemployed who have had regular working activities in the past.¹ A number of them may be working in the irregular economy. As will be explained in Section 8, the footprints of transitions in the irregular economy are difficult to discover, and rough estimates of their magnitude and dimensions can be obtained only through appropriate comparisons across statistical aggregates from different sources. As will be shown, our evidence suggests that once 2–3 years have passed since the beginning of a non-employment spell, only a minority of the long-term non-employed will ever return to a regular working life.

While the EU Labour Force Survey counts unemployment (UN) as those who declare to be unemployed, the OLF (“out-of-the labour-force, but willing to work”), the sum of the two rates (UN + OLF) is suggestive of a “real unemployment rate” of the official economy, with Italy and Spain characterised by the two largest and Germany by the smallest in Europe. Only a fortunate combination of sufficiently similar databases in the three countries allow this analysis.

3. Review of relevant literature

Countless academic studies by economists² investigate the consequences of long-term unemployment: obsolescence of human capital, stigma and perceived signals of ‘bad’ performance, all of which result in wage loss at the time of re-employment. They all refer, however, to ‘long term unemployment’ as defined by official statistics, namely longer than 12 months and only at times longer than 2 years.³ This difference makes them modestly relevant here.

A. Krueger is, to my knowledge, the only economist who has provided an important perspective on the problem of unemployment and non-employment duration. In BPEA [4], he looked at long-term unemployment duration as reported from the CPS monthly data: among workers who answered that they had been unemployed for 27 weeks or more in a given month in the period 2008–2012 – they were re-interviewed 15 months later– 30% were still unemployed and looking for a job, 34% were not working nor looking for work and only 36% were employed. Krueger strongly emphasises the social problems associated with very long non-employment duration: changes in individual lifestyles, family and childbearing projects, increasing poverty and welfare at large. Moreover, he adds “…once a person leaves the labour force, he or she is extremely unlikely to return (at work)”. Unfortunately, Krueger’s methodology does not allow to observe spells of non-employment lasting years therefore, comparisons with LTNE duration as measured in this paper may appear misleading.

Sociologists have paid more attention to the dramatic impact of very long unemployment on lifestyles than economists, although empirical estimates are almost never reported. Newman [5] expresses deep concern over millions of people who became downwardly mobile in the USA between the mid-1960s and the mid-1980s as a result of downsizing, plant closings and mergers. Those who suffered the most were the middle-aged computer executives, the blue-collar workers laid off with the post-industrial economy, the middle managers whose positions were phased out in the same period and housewives stranded with children. Keating [6] discusses the enduring impact of long-term unemployment on developmental health. Brand [7] indicates a decline in psychological and physical well-being, loss of psychosocial assets, social withdrawal, family disruption and lower levels of children’s attainment at school and well-being. Van Horn et al. [8] report the results of a field study on a sample of unemployed men in the USA: among the 13% who had been unemployed for more than 2 years, two-thirds reported a high degree of stress in their family relationships.

4. Longitudinal employer-employee databases

4.1 Italy

The WHIP (Work Histories Italian Panel) longitudinal database originating from Social Security records is a large sample (1,90) representative of the universe of employees in the private sector, the non-tenured employees in the public sector, the self-employed and the professionals as well as all workers covered by atypical (non-standard) contracts.

While prevalent among youths, premature exit takes place at all ages; thus, young non-survivors will no longer be young as time elapses. WHIP covers individual working careers from entry to retirement at monthly frequency, with data on skill level, wages and labour costs, industrial sector, firm size and geographical location, including periods of temporary layoff subsidised by the Cassa Integrazione Guadagni. Instead, it does not identify unemployed individuals not entitled to draw benefits. Data on educational attainment are unrecorded in the WHIP database.

4.2 Spain

The Spanish labour market is studied by means of the administrative MCVL (Muestra Continua de Vidas Laborales) database that covers all workers, whose first spell of employment took place when they were between 16 and 30 years of age. MCVL is a representative longitudinal sample of the population registered with the Social Security Administration. The raw data represent a 4% random sample of the reference population (pension earners, unemployment benefit recipients, employees and self-employed workers), amounting to approximately 1.2 million individuals each year. Self-employed are included in MCVL. The same holds for public employees as long as they contribute to public social security (some categories report to their own social security administration). MCVL offers retrospective information, that is, the entire labour history of the workers registered with the Social Security Administration during the year the sample is extracted.

4.3 Germany

The analysis on Germany is performed using the SIAB database. SIAB (Stichprobe der Integrierten Arbeitsmarktbiografien/Sample of Labour Market Biographies) is supplied by the IAB (Institut für Arbeitsmarkt- und Berufsforschung/Institute for Employment Research). The BA administers the German unemployment insurance and therefore has access to social insurance records as well as unemployment benefit data and labour market policy programme data. The SIAB covers the data with a sample/population ratio of 2%, where the population consists of people registered as dependent employees or recipients of unemployment benefits/participants in labour market policy programmes at some point in time between 1975 and 2014 (West Germany, East Germany: 1992–2014; see [9]).

5. The measurement of survival

5.1 Methodology

The basic statistic used in this exploration is labour market “survival”. Notice that our use of the term “survival” differs from the standard one found in statistical literature. We estimate the number of “labour market survivors” by counting the individuals employed since a given starting year and still at work at the end of the observation period, even if they have had periods in unemployment in the course of their careers. Thus, the non-survivors are individuals who disappear from the database before the end of the observation period and no longer reappear.⁴ If anyone is unobservable for a period of time and then reappears as re-employed, that period is considered a spell of non-employment. Such spells may last months or years (additional schooling is, obviously, a likely possibility for young men), but they ultimately lead to re-entry into employment or self-employment. Spells in registered/official unemployment and paid sickness periods are observed in the administrative data and not counted as periods of absence from the labour market.

Figure 1 exemplifies the counting method of a cohort of 8 individuals, A – H, whose work histories are observed between 1987 (the year of entry for all) and 2008. D and E drop out of the labour market two years after entry (but D will re-enter one year later). B leaves his post in 1991, finds a new job and loses it again in1997, is back at work in 1999, but shortly afterwards leaves the job market and no longer re-enters. If the observation window were to end in 1989, survival would be 6/8 = 0.75. In 2001, the non-survivors are A, E and F, and the 2001 survival is 5/8 = 0.625. In 2008, the only survivors are D and H, all the others having dropped out before 2008 (leaving survival at 2/8 = 0.25).

Figure 1.

Example of counting survival (continuous lines denote employment spells).

Censoring at the end of the observation period may be the cause of upward or downward bias of survival. Ad hoc unbiasing procedures are applied (see [10]).

5.2 Some notes of caution on comparative analyses of survival

Caution is due when comparing survivals from different countries and contexts. Firstly, the survival rates depend on the length of the observation period, and its timing may be affected by the business cycle [9].

Second, entrants must be in the same age range: age has an impact on survival, with the younger entrants surviving longer than their older peers.

Third and last, but not least, not all administrative databases cover the same forms of working activities: some include all dependent employees but leave out self-employment and civil servants (Germany); others include self-employment but leave out seasonal workers in the agricultural sector and tenured public employees (Italy); others may exclude certain contract typologies. These are problems that must be handled ad hoc, integrating our administrative data with different databases, as discussed in what follows.

6. Survival: a survey of results

Analysis of survival is performed on cells defined by cohorts of young male entrants observed at one-year intervals. While the overall observation period 1980–2012 is available for all countries, detailed information on the workers’ and employers’ characteristics is not. Considering the longest available period (1980–2012) for domestic nationals, Italy’s 32-year survival rate is 80%, the West German rate stands at 95% and the Spanish rate is in the middle (88%) between the former two, despite very high Spanish unemployment in 2012. Transitions into self-employment play an important role in preserving a high level of survival after periods of non-employment. It explains about 11 pp. of survival in Germany, more than 20 pp. in Spain and about 18 pp. in Italy. The age of Spanish new entrants is set at 16, lower than either of the other countries under study. This explains to some degree the higher survival rate compared to Italy, while that of Germany is a likely consequence of more efficient procedures to re-employ job losers.

6.1 Subgroup analyses of survival

The first observable year of entry with complete subgroup information is 1987 in Italy, 1993 in reunified Germany and 1991 in Spain. Our subgroup analyses are based on these starting years. Also, the entrants’ age with complete information is not the same: 19–30 in Italy and 16–30 in Germany and Spain (Table 1). The following dimensions are objects of inquiry:

• age group of the relevant cohort.

• year of first entry into the labour market.

• duration of first employment spell.

• economic branch of initial activity.

• geographical area.

• size of first employer’s business.

• mobility (movers vs. stayers).

• skill level.

• education level.

• entry wage.

	Unemployment UN	Out-of-labour force OLF	UN + OLF	Youth UN
Italy	12.9	10.8	23.7	42.7
France	10.4	2.8	13.2	24.2
Germany	5.1	2.6	7.7	7.7
UK	6.3	4.8	11.1	17.0
Spain	24.2	.5.0	29.6	53.2

Table 1.

Unemployment and out of labour force (% rates) in 2014.

Table 2 displays the 2012 average survival rates for all the above dimensions. It.is interesting to notice that the main features of survival are similar across all three countries.

1. Italy’s 25-year survival rate (1987–2012) is 79%. Survival among individuals who become self-employed in the course of their career is slightly higher.

2. The 21-year survival rate in Spain (1991–2012) is 87%, considerably higher than Italy’s rate, despite unemployment reaching almost 25% in 2012. Here too, frequent moves into self-employment after 2006 explain such high survival. The age of Spanish new entrants is 16, lower than either of the other countries.

3. The 19-year survival rate in Germany (1993–2012) stands at 95%, considerably higher than the Spanish and the Italian rates.

4. The impact of age at the time of entry is similar in Italy and Spain, with young entrants surviving longer than their older colleagues: the difference is more pronounced in Italy (21 pp. between the youngest and oldest cohort) than Spain (12 pp.), while German older entrants survive more often (5 pp.), presumably because they often start as civil servants.

5. The length of the first employment spell provides an indication of how employers evaluate the ability of the prospective recruits. It is reasonable to assume that employers who interview a promising young person will offer them a longer starting contract than a less interesting candidate. In Italy, the survival of workers starting in 1987 with a long initial spell of employment (12 months +) is about 87%, whereas it drops to 68% for those whose first employment spell lasts less than 3 months. The latter are characterised by an abrupt drop of survival in (t + 1) and (t + 2), followed by a slow decline thereafter. A similar pattern is found in Germany. In Spain, the differences are less pronounced as many very short starting spells have not been recorded.

6. An additional indicator of individual ability as it is appraised by the employers is the starting salary: a promising worker will presumably be offered a higher wage than a less promising one, and his survival is likely to be higher for the same reasons indicated above. The differential survival between workers (here, the blue collars) with starting salary in the upper quartile (Q4) of the distribution and those in the lowest quartile (Q1) cum a short initial spell (< 3 months) is remarkable in Italy and Spain (German data are unavailable). This finding strongly indicates the value attributed to human capital by the employers. Overall, bad starts have a strong and persistent effect on future labour market outcomes, even when the future lies 15–20 years ahead. The ECHP exploration (described in Section 9) suggests that the non-survivors are likely to be the least endowed also in terms of education and family background.

7. The education level is observable in Spain and Germany, although the reported degrees are not the same. In Italy, schooling is not recorded in the administrative databases. Education has a predictable positive impact on survival, but differences are small (3–6 pp.). Evidence from the ECHP database (Section 9) indicates that educational differences also have a similar impact in Italy.

8. The impact of mobility on survival is very important (geographical, as well as job-to-job, often with intervening unemployment spells between job switches). Workers who perceive their job to be at risk start searching for more solid positions, and many appear to be successful. In Germany and Italy, the stayers (no moves) are shown separately and display a much lower survival than their moving colleagues: 61 and 49%, respectively. Individuals who have moved up to 5 times in their career survive much longer everywhere. Very frequent movers (10 + moves) survive even longer.

9. The employers’ location reflects to some degree the different degrees of industrialisation and regional wealth. The survival differential is marginal in Germany (the Länder of the West are the rich ones compared to those of the East) and Spain, while the divide is larger in Italy between North and South.

10. The qualification divide is also important, in line with the idea that human capital makes a difference: white-collar workers survive somewhat longer than blue-collar workers everywhere: 95 vs. 92% in Germany, 91 vs. 86% in Spain and 80 vs. 77% in Italy.

11. In this respect, it would have been interesting to have more detailed data of the occupational profiles, but for the time being, they are not available.

12. The size of the employer’s firm leads to interesting insights. In Italy, the divide is between “smaller size = <20 employees” and “larger size = 200+ employees”. In Spain, the divide is “small = 26-50” and “large = 50+”. Here we find noticeable differences in the expected direction of longer survival among the larger firms. It is small in Spain where a more incisive definition of “large” would have been helpful. Lifetime employment in the same establishment was frequent many years ago, but no longer nowadays. There is a vast body of economic and sociological literature indicating that higher work attachment and loyalty prevail in small establishments where a variety of on-the-job duties instead of repetitive tasks, more opportunities for expanding and upgrading knowhow and personal ties established with peers and employers are frequently offered. There is also empirical evidence that small firms place a higher value on human capital than large enterprises. This, however, may not translate into longer survival as small firms are more exposed to the turbulence of economic life, when turnover is high and when closures and bankruptcies are frequent events. This seems to be the case in Italy and in Spain. In Germany, an average employer’s size is much larger than in Italy and Spain: the breakdown <200 workers (smaller) and 200+ workers (high) cannot be easily compared to the one meaningful in Italy and Spain. It may not be surprising that survival in German “smaller” firms is higher than among the “larger” ones.

13. The survival rates computed in correspondence with the business cycle prevailing at the time of each cohort’s entry appear meaningful at first sight (larger at the high cycle, smaller at the low cycle). The cyclical impact, however, could be more ambiguous than it appears. New entrants during the low cycle could be chosen among highly skilled workers in order to recoup productivity delays, while those hired in expansionary years could be temporary workers aimed at filling short-time vacancies. This will positively affect survival among the former and reduce it among the latter.

14. The qualification divide is also important, in line with the idea that human capital makes a difference: white-collar workers survive somewhat longer than blue-collar workers everywhere: 95 vs. 92% in Germany, 91 vs. 86% in Spain, 80 vs. 77% in Italy. In this respect, it would have been interesting to have more detailed data of the occupational profiles, but for the time being, they are not available in the administrative databases.

	Italy 1987–2012	Germany 1993–2012	Spain 1991–2012
Overall survival	79	95	87
Age at entry
19–21	81	95	88
26–30	60	100	76
First job duration
Short (<6 months)	68	90	86
Long (12 months +)	87	98	91
Initial wage quartile
Q1 and job spell<3 m.	71	n.a.	79
Q4	82	n.a.	92
Education degree
Compulsory school	n.a.	92	84
Higher education	n.a.	95	90
Mobility
0 moves = stayers	40	61	n.a.
3–5 moves	78	91	86
10 + moves	93	100	87
Regional wealth
Low	72	94	87
High	82	99	87
Skill qualification
Blue	77	92	86
White	80	95	91
Employer size
Small	65	99	85
Large	78	94	86
Business cycle
Low	76	83	91
High	81	96	98

Table 2.

Average survival rates.

Source: our calculations.

7. Long term non-employment duration

Estimates of LTNE duration are obtained with a simple ad hoc procedure: individuals of each cohort are counted in employment month after month during their working life. Once they drop out and no longer reappear anywhere in the labour market, they are counted as non-employed, and their LTNE duration is measured from that moment to the end of the observation window.

The full observation window spans between 1987 and 2012 in Italy⁵, 1991 and 2012 in Spain and 1993 and 2012 in Germany, covering all the male cohorts entering the labour market during that time window. The LTNE count, therefore, includes all individuals still of working age at the end of 2012.

People aged 30 in 1980 will be 62 in 2012. The upper limit of the oldest age bracket may reach 66 as retirement age is not identical in Italy, Germany and Spain and has been the object of changes during the observation window.

Tables 3 and 4 display: (i) the estimated magnitude of uninterrupted long-term non-employment ending in 2012; (ii) the average LTNE duration; (iii) the ratio of LTNE individuals to the male population of working age (MPWA). A good assessment of the magnitude of LTNE is provided by its ratio to the male population of working age (16–64), rather than to total workforce that may have margins of ambiguity.⁶ Not surprisingly, in Germany, this ratio is 2.9%, much lower than 6.5% in Italy and 6.1% in Spain. If unemployment rates were calculated as a percent of the same denominator MPWA, the comparison becomes even more transparent: 3.4% in Germany, 7.2% in Italy and 19.9% in Spain (while the official 2012 male unemployment rates were 5.7, 10.0, and 24.7%, respectively).

Age	LTNE duration (yrs)	Italy	Germany	Spain
53++	25–32	7	12	0
47–53	21–24	8	8	1
38–46	16–20	21	17	13
32–37	10–15	28	20	21
26–31	5–9	32	23	26
16–25	0–4	3	20	40

Table 3.

Long-term non-employment magnitude (000) and duration in 2012 the share of LTNE in each age group.

Source: own calculations.

	OLF	LTNE	Avg.LTNE (years)	LTNE/ MPWA	UN / MPWA	Official UN rate
Italy	1421	1260	11.6	8.5%	7.2%	10.0%
Germany	589	756	12.8	2.9%	3.4%	.5.7%
Spain	499	973	6.2	6,1%	19.9%	24.7%

Table 4.

LTNE and “out of the labor force but willing to take a job if available”” (OLF) magnitude (000), average duration and share in male working age population 2012 (MPWA: Age 15–64).

Source: Eurostat based on LFS, n calculations (LTNE, see section 6).

The macroeconomic determinants of LTNE are similar to those that impact unemployment and long-term unemployment, the main one being the lack of aggregate demand. In addition, as already pointed out, unemployment insurance arrangements have an important impact on the LTNE magnitude. Where they are generous as in Germany, and to a lesser extent in Spain, dismissed workers have less incentive to withdraw from the labour force and become LTNE as long as they can draw unemployment benefits (up to 24 months in Germany and 12 months in Spain). In Italy instead, the incentive for many jobless people to apply for unemployment benefits is modest as the Placement Centres are not as efficient as elsewhere. Consequently, the probability of falling into the LTNE is higher.

The black/irregular economy is an important safety valve available to the jobless, not only in the course of recessionary periods: in fact, there is no evidence that it is anti-cyclical. It is especially present in Italy and Spain, while it is all but non-existent in Germany. In the irregular economy, pay is often modest and social protection weak, seldom providing a shelter to irregular workers and their families.⁷ But the mere possibility to do some work cum tax evasion are huge incentives for many to enter the irregular sector. And it helps also regularly employed people who may want to round up their budgets. Schneider estimates that income generated in the irregular economy varies among 13% in Germany, 18% in Spain, and over 20% in Italy.

Noticeable country differences are found in the shares of each age group. In Germany, the large proportion of older individuals (53++) is attributable to the large inflow of East German workers during the years of reunification. In Spain, instead, the employment increase that took place with the reforms introduced in the period 2000–2007 explains the greater presence of young cohorts.⁸ The age group (57–66) is relatively small as many individuals have retired before the end of the observation period. The 32–46 age groups are very numerous, with average LTNE durations of 10–20 years. Such a duration is indeed dramatic as all these people are prime-age adults who spend most of their life outside the labour market.

Average LTNE duration of the youngest group (16–25) is 0–4 years. In Spain, 40% of all LTNE individuals belong to this age group, as entry age is recorded since age 16. Elsewhere entry age is 19, and therefore, the numerosity of the same age group is lower.

The number of male OLFs is estimated by the European Labour Force Survey. As explained in part 2, it would not be surprising to find them of the same order of magnitude as the LTNEs. This is the case for Italy and Germany, while the figure for Spain is about one half the LTNE estimate.

An important question relates to the personal characteristics of the individuals who are more likely to become LTNEs. Administrative data contain few answers. For this reason, we match our administrative data with the ECHP survey, as reported in part 9.

8. The big question: where do the LTNEs end up?

The crucial question is where all the long-time jobless individuals end up. The shadow economy is an obvious candidate as an end destination. Discovering it, however, is a difficult task as no micro-data of irregular work are available to help with the answer. Current measurements of the irregular economy are based on rough macroeconomic indicators.⁹

Schneider and Ernste [11] estimated the share of irregular activities on GNP for several OECD countries: Italy ranks the highest at 21.5%; Spain’s share is 19.2%; Germany’s is among the lower ones with 13.5%. The share of irregular employment in overall employment in Italy and Spain is estimated at around 16% and at 12.5% in Germany. It may seem odd that the share of irregular employment is lower than the share of shadow GNP, given that the productivity of irregular workers must be lower than that of regular workers. The answer is that the very high income generated by outright criminal activities often gets laundered in regular production activities ending up in GNP.

There are no micro-based databases to uncover the economic and social backgrounds of irregular workers, nor what happened during their working career that encouraged them to join the irregular economy. Limited information can be obtained from LFS-type surveys. In the next section (9), we report some encouraging results of a comparative analysis performed on ECHP data.

Table 5 summarises data from different sources, including Schneider’s estimates of the size of the EU shadow/irregular economy for the year 2012. Interestingly, the LTNE estimates of Germany and especially Spain are remarkably close to the available estimates of irregular workers; less so for Italy.

	Unemployed men in 2012	LTNE	Irregular male workers	Schneider’s irregular economy as % of GNP ^
Italy	1766	1260	1800 (**)	21.5%
Germany	1505	756	950 (*)	13.3%
Spain	2957	973	968 (***)	19.2%

Table 5.

LTNE and the irregular economy (000).

Sources: (*) IAB estimate; (**) ISTAT National Accounts, (***) estimate based on Schneider and Enste [11]; (^) Schneider and Enste [11].

9. Benchmarking worker disposal with the ECHP 1994–2002

A natural benchmark of the estimates of survival is provided by the European Community Household Panel (ECHP), observed between 1994 and 2002 (the choice for the older ECHP rather than EU SILC that replaced it in 2002 is dictated by its longer longitudinal structure, eight years vs. four years of the latter). The ECHP survey provides information on a number of personal characteristics, general income conditions and work contract typology that are absent from the administrative databases.

We estimate survival in the ECHP data as we have done with the administrative database, selecting the individuals who self-report as working until year (t) and not after. Once no longer working, but still responding to the ECHP questionnaire, they report either as unemployed or inactive. Their status is similar to the individuals whom we define as LTNE, the main difference being that in the ECHP survey, they report their status after the last job termination, in addition to some of the circumstances that lead to joblessness. No explicit indications are present on shifts into the black economy, although there are occasional hints to this effect.

We briefly summarise here the relevant information. LTNE people appear to be worse off than the other respondents on almost all counts: higher unemployment experience, higher job search activity, lower family income, frequent elementary occupations, more difficulty to make ends meet and lower educational degree. The share of people who skip the answer to the contract typology is high, almost one-third of all sampled individuals in Italy and Germany and about one-sixth in Spain: the choice of not answering this item could hide some presence in the irregular economy. Some respondents report to have worked in the absence of any contract, another hint of activity in the irregular economy. It should be noted that irregular workers may also self-report as regularly employed for fear of being discovered (in which case, we would have no hints to help their recognition). In all three countries, the individuals who are likely to have become LTNEs report low educational degrees and few skills.

The condition reported in the ECHP after premature exit indicates unemployment for two-thirds of the people in Italy and Germany and almost half in Spain. Exit could be the consequence of quitting or of involuntary dismissal. Voluntary quitting (family reasons, study, military career, better opportunities) is reported to be very high in Italy (60% of answers) and very low in Spain (9%), with Germany midway. While the data from Germany and Spain appear coherent with other answers, those from Italy do not. In Germany, many report voluntarily quitting, and many, coherently, declare to be inactive after their last job; in Spain, the high number of involuntary job losses matches the high frequency of unemployed. In Italy, instead, we see several voluntarily quitting cum few inactive. A plausible explanation may reside in a common yet illegal practice followed (especially in the past) by many employers in order to avoid the firing costs associated with unjustified layoffs: at the time of a new hire, the worker was requested to sign a letter of voluntary resignation held by the employer. Many newly hired individuals would agree for fear of losing the job. If the employer decided to layoff for whatever reason, the letter would serve to show that it was the employee’s voluntary decision to terminate his engagement, and no firing costs could be levied on the employer.

10. Policy indications

The dilemma of how to effectively deal with youth unemployment is present also in this study of LTNE. Past policy in almost all EU countries was prevalently supply-sided, focused on enhancing contract flexibility and cutting labour costs through subsidies to the employers. Active labour market policies focused instead on setting up training facilities for the young and retraining and upskilling schemes for the adult, long-term unemployed. All in all, this approach performed poorly in the aftermath of the 2008 crisis especially in Italy and Spain, although serious problems of youth unemployment have been present elsewhere in the EU since the turn of the millennium.

While we strongly envisage the need to turn to demand-side policies capable to have a direct impact on employment, our explorations suggest also a number of supply-side implications, not applying equally to the three countries under observation:

1. A very general indication, valid especially for Italy and, to some extent, for Spain, is the need to improve the match between demand for higher skills and supply, by investing in the education system and strengthening the placement and re-training agencies (public and private). Excessive worker turnover ¹⁰ frequently leading to market dropout hints at problems of unsatisfactory matching. The share of expenditure in active labour market policies (ALMP) in GNP is very low in Spain (0.8%) and especially in Italy (0.6%) compared to Germany (1.1%).

2. We see no reason for introducing additional flexibility to contract termination. Measures aimed at facilitating the transition between precarious jobs and permanent positions are instead very desirable (also in Germany, where the dropout rate after termination of mini-jobs appears to be quite high).

3. Nor do we see compelling evidence that measures aimed at further reductions of labour costs (mainly by way of tax subsidies) would substantially improve youth employment. As of today, the labour cost of young people is already lower than that of their older working counterparts (in Italy, youth earnings of unskilled blue-collars are about 60% of adult earnings with similar skills). The incentive to upgrade human capital is lost if the cost of hiring a new recruit is too low compared to the cost of retaining and upgrading young workers already on the job. Moreover, the current practice of high turnover, i.e. replacing people already on the job with new unskilled recruits, generates unwanted consequences of premature exit from the labour market and ultimately on long-term non-employment.

Some implications apply specifically to the Italian case:

1. It is crucial to improve the generosity of unemployment benefits. Initial steps in this direction were taken with the Fornero reform in 2011, but almost none has been as yet implemented. In Italy, only a small share of Italy’s working population is eligible for unemployment benefits: Italy’s recipiency rate is 32%, against 50% in the UK, 60% in France, 65% in Denmark, 73% in Spain, 94% in Austria and 100% in Germany, although these rates do not imply the same degree of generosity. As a consequence, the Italian unemployed/non-employed workers have little incentive to self-report to the Placement Centres because the opportunity cost is often close to zero. Where unemployment benefits are generously available, as in Germany and, to a lesser degree, also Spain, the opportunity cost of misreporting is high because the perceived risk of losing the benefits is equally high.

2. Less invasive regulation is necessary. A large number of jobs, perfectly legal in many EU countries, are ‘irregular’ by Italian standards: many low-paid, often part-time or temporary jobs in the service sectors, such as waiters, janitors, salespeople, domestic helpers and caretakers, are held mainly (but not exclusively) by young people. A reform on this terrain would restore the incentive to work in the ‘regular’ economy, enjoying the benefits of social security and at the same time paying very modest taxes.

Demographic trends in the coming decades are likely to improve the job prospects for younger generations: the baby-boomers will begin to retire in the next decade, and their replacement ought to increase the demand for young workers. A major labour shortage may be around the corner in Europe. And it will spur additional massive migrations of largely unskilled migrants from non-EU countries with high fertility rates. This will be a source of ever-growing governance problems for the European Union, as social unrest will not cease to lurk outside the door.

11. Conclusions

This study proposes a new approach to the analysis of long-term non-employment and its duration in Germany, Italy and Spain using administrative longitudinal databases. The non-employed are people of working age who have had spells of regular employment at the beginning of their careers, were laid off and became unemployed, but no longer succeed to regain a job in the official economy. For many, non-employment turns out to be a lifetime condition. Its duration may reach 25–30 years. The number of the long-term non-employed (LTNE) is about the same order of magnitude as official unemployment.

A comparison of LTNEs with official labour market aggregates and estimates of the irregular economy suggests that a large share of the LTNEs may join the irregular employment. Benchmarking the LTNE estimates from administrative data with EHCP survey allows to trace some individual characteristics unavailable in the administrative data, the main ones indicating that the probability of becoming LTNEs is higher for people with a low educational background and low skills.

During the study period, a number of changes took place in the economies of the three countries. They have been taken in due consideration: nonetheless some conclusions should be taken with care.

Additional information

An earlier version of this chapter was previously published as a preprint in: New Approaches to the Study of Long-Term Non-Employment Duration in Italy, Germany and Spain [Internet]. www.iza.org. [cited 2023 Nov 19]. Available from: https://www.iza.org/publications/dp/11167/new-approaches-to-the-study-of-long-term-non-employment-duration-in-italy-germany-and-spain