Have labour market reforms at the turn of the millennium changed job durations of the new entrants?

Transcription

1 Have labour market reforms at the turn of the millennium changed job durations of the new entrants? Gianna C. Giannelli (Università di Firenze) Ursula Jaenichen (IAB, Norimberga) Claudia Villosio (LABOR, Torino) Conference on: Regulating for Decent Work ILO, Geneve, 8-10 July 2009 July 2009

2 Focus The analysis focuses on the careers of young people entering the labour market during a period of Labour Market Reforms (LMR)

3 Motivation General framework: trade-off between stability and employment opportunities for new entrants First question: has job stability of new entrants decreased with labour market reforms? Second question: have LMR increased employment opportunities? Methodology: flexibility/opportunities measured with durations Comparison: two regimes in the 90s 1) smooth reforms (Germany) 2) radical reforms (Italy);

4 Steps First step: assessing flexibility. Test the hypothesis that first job durations have decreased during LMR period. Analysis of first job durations Second step: assessing job opportunities. Test the hypothesis that if the first job lasts less, then it is rapidly followed by other jobs. Analysis of employment durations Job vs. employment duration

5 Literature on job stability Use of duration to measure job stability, Booth at al. (1999), separation hazards were higher for more recent cohorts, implying a secular increase in job instability Vast literature on fixed term contracts as stepping stone into permanent contracts; see e.g. Gagliarducci (2005); Berton, F., F. Devicienti and L. Pacelli (2007); Boockmann, B., and T. Hagen (2008) Not much literature job duration and reforms: see e.g. Garibaldi and Pacelli (2008), Bonnal et al. (1997) We aim at contributing to the literature on reforms and job stability

6 Is there a clear time pattern of changes in job durations along with the timing of the reforms?

7 Institutional background Italy and Germany: ease their employment protection legislation during the 90s and the early 2000s Different intensity and pace of these reforms: continuous and moderate as well as partly reversed in Germany, and once for all and drastic in Italy.

8 Kohl Government (until 1998; deregulation ) 1985/1989/1994/1996/1998: - stepwise facilitation of fixed-term contracts and temporary agency work - dismissal protection law (1996) 1998: Social Code III - new role of active labour market policy - more responsability for unemployed persons - stepwise reduction of public job creation in East Germany - massive use of private wage subsidies - innovative active labour market instruments 3

9 Schröder Government ( ; reregulation ) 1999/2001: reregulation - dismissal protection law - fixed-term and part-time contracts - new regulation of marginal employment 2001/2002: Job-AQTIV law confirms activating strategies in the orientation of active labour market policy 2002/2005: Hartz reforms/agenda restructuring of federal employment services - reform of active labour market policies/new instruments - reform of unemployment and social assistance 3

10 Bottom line for reforms in Germany in the 90 s rather smooth reforms with respect to more flexible working contracts partly trends towards re-regulation plausible hypothesis: reforms have the potential to ease labour market entry for younger persons at the cost of providing them with less job stability for Germany rather small effects expected

11 Italy Reforms in Italy in the 90 s 1995 law 335/1995: extension of compulsory social security to parasubordinati (co.co.co and professionals) 1997 Treu s law: introduction of temporary work agency 2001 Deregulation of fixed term contract 2003 Biagi s law: incentive to hire employees for shorter periods 2

12 Macroeconomy Both in Germany and Italy in the nineties: - high unemployment and growing long-term unemployment (extremely high unemployment rates in East Germany) - recession 1993 (Germany after reunification boom) - modest upswing 1996/

13 German data: administrative IABS Representative 2-percent sample of employed persons Employment only if subject to social security contributions: no self-employed, no civil servants with status Beamte Marginal employment (now: mini-jobs) only since 1999 Spell data based on employers notifications (at least once a year, employer change, changes in working contract,...) Data on benefit receipt: neglected for the moment

14 Italian data: administrative WHIP Representative sample of employed persons (sampling ratio 1:90) Covers all the people Italian and foreign who have worked in Italy even for only a part of their working career Employment only if subject to social security contributions: no self-employed, no civil servant and professionals with autonomous found excluded The observed period goes from 1985 to

15 The econometric model: Piecewise constant proportional hazard λ j ( β λ β t x ) = ( t)exp( x ) (1) λ t) = λ 0 j i 0 ( with τ < t < τ j 1 j (2) i The hazard rate may vary over time (duration dependence), but is constant within intervals the covariate effects are constant over time

16 Sampling Flow-sampling of new entrants: employees recorded for the first time in the archive, aged We observe them in a 3 year window Germany, Italy No left censoring No information is available on earlier preemployment periods

17 Definition of first job regular employment - no marginal jobs (not available for all years) - no vocational training first employment with highest level of education - possibility of moving between educational system and the labour market 4

18 Graph 1 Number of new entrants into dependent employment by sex, Germany,

19 Number of entries, descriptive results, Germany roughly first job spells more entries since 1996 (upswing, total employment still going down) the majority of entries still occurs after vocational training in private firms about 3000 yearly job entries in East Germany, no upward trend in the late 90s like in the West more men than women strong seasonal pattern in the number of job entries mean age at job entry in our data is fairly constant at nearly 24 years 4

20 Graph 2 Number of new entrants into dependent employment by sex, Italy, female male

21 Number of entries, descriptive results, Italy roughly 7000 entries on average each year more entries since 1993 more men than women (==Germany) strong seasonal pattern in the number of job entries (==Germany) Average age at entry slightly increasing over time from 22 in 1994 to nearly 25 in 2002 Immigrant workers: a peak in corrispondence of two important regularization laws 4

22 Durations measured in days, Germany, or months, Italy Descriptives are in months artificially censored after 3 years censored if the spell ends in the last year of our observation window A spell is considered as continuos if there are interruptions without employer change (maternity leave, sickness,...) up to 6 months

23 Graph 3 First job survivor functions, Germany

24 First job duration, descriptive results, Germany 50 percent of first jobs end within the first 12 months at the end of th 3-years window, about 25 percent of all first jobs are still going on for East and West Germany, there is a slight downward shift in the survivor functions, comparing the entry years and men have lower survival rates than women for men in West Germany there is no clear downward trend in the survivor function over time the survival rates of unskilled workers are much lower than those of (the larger group of ) skilled workers 4

25 Graph 4 First job survivor functions, entry in 1990 and in 2000, Italy

26 First job duration, descriptive results, Italy More than 50 percent of first jobs end within the first 12 months (54% in 1990; 60% in 2000) (== Germany) At the end of th 3-years window, 17 percent of all first jobs are still going on (< Germany) There is a downward shift in the survivor functions, comparing the entry years 1990 and 2000 (== Germany) Men have lower survival rates than women but the drop in duration is higher for women (== Germany) The survival rates of unskilled workers (blue collars) were much lower than those of skilled workers (white collars), but skilled workers suffer more of the decline in duration. 4

27 Modelling first job durations For both countries we estimate the same model Dummies for year of entry to capture the influence of labour market reforms Control variables: ITALY -seasonal dummies - local unemployment rate - local value added growth -region - firm size - industry -foreign -age - skill - part-time -CFL and agency contract GERMANY -seasonal dummies - local unemployment rate local value added growth - region (Bundesländer) - firm size - industry -foreign -age - education/skill - part-time 3

28 Table 3 First job durations: Hazard ratios of the year dummies (Germany: base year 1994, Italy: base year 1990) GERMANY ITALY MALES FEMALES MALES FEMALES significant at α=0.05 or less

29 Interpretation of estimated coefficients The YEAR DUMMIES: In both countries increase in the hazard of quitting. This process corresponds to some LMR Differences: In Germany, increases concern women. In Italy, increases affect both sexes Duration dependence: Strong and negative, the risk of leaving the first job decreases for longer tenures 4

30 First job duration. Estimation results. Germany, control variables - significant and strong effects of seasonal dummies - significant positive effect of the local unemployment rate - mostly significant effects of the Länder dummies - GDP growth decreases job durations - significant and strong effects of firm size: job durations are longer in larger firms for both men and women - significant effects of industry, stronger for men - significantly shorter durations for non-germans - significantly longer durations for higher entry ages - strong and significant effects of training and education, job durations are rising with skill level - significant effect of part-time: part-timers have shorter first job tenures 4

31 Duration of first job. Estimation results. Italy, control variables - significantly longer durations for higher entry ages (==G) - significantly shorter durations for immigrants ONLY for female - job durations are higher for apprentices (due to the nature of contract), CFL, part-timers ( G) and white collars; lower for blue collars and agency workers - significant and strong effects of firm size: job durations are longer in larger firms expecially for men (= G) - significant effect of the local unemployment rate (negative sign: G insider theory ) - positive effect of demand (VA) - significant and strong effects of seasonal dummies - significant effects of regions (lower durations in the south) 4 - significant effects of industry, stronger for men (==G)

32 Work careers: employment and unemployment

33 Job mobility What happens when the first job ends?

34 Table 4: Distribution of new entrants by number of jobs held in the first 3 years after entry and by year of entry (levels and %) GERMANY Year of entry or more Total Total

35 Table 4 (continued): Distribution of new entrants by number of jobs held in the first 3 years after entry and by year of entry (levels and %) ITALY Year of entry or more Total

36 Number of jobs over the 3 years In Germany, the share of persons with only one job spell goes down from 49 % for the 1994 entrants to 42 % for the 1999 entrants. In Italy, the number of people with only one job drops from 57% in 1990, to 48% in 2001, with a low of 42% in Polarization in both countries: one-job share decreases, more than 2 jobs-share increases at the beginning the share of one-job Italians larger than one-job Germans, while at the end more similar. 4

37 First Job vs Employment Durations Definition of continuous employment: formed either by only one job spell or by different job spells with a different employer, with a maximum interruption of three months between them.

38 Trade-off between stability and employment opportunities : EPL FJD Job mobility faster job matches stable ED possibly detrimental effects of multiple (short) job spells on the accumulation of human capital not considered in this analysis

39 In Germany mobility between jobs is quite large 40% of the employment spells are composed by more than one job, and 20% by more than two jobs. In Italy the share of one-job spells is much higher than in Germany, around 75% on average, and remains fairly stable over the whole period. Lower degree of job mobility in Italy The results for the estimated duration model will not change dramatically when switching from the job spell to the employment spell concept.

40 Table 6 First employment duration: Hazard ratios of the year dummies (Germany: base year 1994, Italy: base year 1990) GERMANY ITALY MALES FEMALES MALES FEMALES significant at α=0.05 or less

41 Summary for the effects of yearly dummies in the employment duration model Germany Males: reject the hypothesis of a decrease in employment duration during LMR; the opportunity to switch rapidly from one job to the other has increased Females: employment duration not increased. So, under this respect, the reforms might be thought to be not completely successful. Italy The reduction in the first job duration has not been counterbalanced by an increase in the opportunity to find rapidly another (or more than one) and, may be, more stable job. This is true for both sexes, for all years, also in correspondence of important labour market reforms.

42 Conclusions The data allow for comparable analysis for Italy and Germany There is a trend to shorter durations in the first job, stronger in Italy than in Germany in correspondence of LMR Employment duration increases in Germany (for males) and decreases in Italy This suggests that too radical changes from too much rigidity to too much flexibility might not yield the expected outcomes.

43 Table A2 - Germany: Estimated coefficients for variants of employment duration models Employment duration (attention: in the estimations, STATA always drops me some 50 persons without comment -could there be remaining data problems??) Males, West Germany Null model without unobserved heterogeneity Haz. ratioz Haz. ratioz Haz. ratioz Haz. ratioz Duration 0-31 days l days l days l days l days l days l days l days l days and more l Number of jobs spells in first employment spell 2-3 jobs jobs or more jobs jobs4m Year of entry 1995 y y y y y y y

44 Quarter of entry apr-jun quart jul-sep quart oct-dec quart Local labour demand (district level) unemployment rate alq gdp growth growth Federal state Schleswig-Holstein, Hamburg bul Niedersachsen, Bremen bul Hessen bul Rheinland-Pfalz, Saarland bul Baden-Wuerttemberg bul Bayern bul Foreigner foreign Age age age age age age age age age Skill level no information on educational level sk no vocational training with at most insk Abitur/equivalent; with or without vosk University/Technical/Professional Cosk

45 Part-time (min. 18h/week) wt Firm size (1st job) fs fs fs and more fs Industry (1st job) agriculture, mining ind energy, traffic and information ind manufacturing ind construction ind trade and retail ind personal and domestic services ind social and public services ind Frailty term (log) ln_theta Reference categories: German nationality, age 20-24, vocational training with at most intermediate education, fulltime, firmsize less than 20, business services. Models with number of jobs in 1st employment: 1 job

46 How to tackle the problem of selection?

47 1) First employment censored after 3 years 1st employment begin 1st employment end of 3 year's window 2) Complete spells for 1st employment and 1st unemployment 1st employment 2nd employment 1st unemployment 3) Complete spells for 1st employment and censored spell for 1st unemployment 1st employment 1st unemployment end of 3 year's window

48