This paper aims to analyze labor market dynamics in Italy by examining inflows and outflows using Compulsory Communications Data on over 263,000 jobs between 2008-2010. The authors build monthly and average annual transition matrices between employment states and examine determinants of job activation and termination probabilities. Key findings include evidence of labor market flows between different contract types and underestimation of terminations due to excluded pre-2008 contracts.

1. Inflows and Outflows in the Italian Labour Market
F. Chelli, C. Gigliarano, M. Lilla, S. Staffolani
Department of Economics
Universita Politecnica delle Marche, Ancona
Enhancement and Social Responsibility of Official Statistics,
Roma, 28-29 Aprile 2011

2. Aim of this paper
• to focus on the central question whether there is sufficient
labour market dynamics in Italy
• to provide an up-to-date measure of the inflows and (partial)
outflows in the Italian labor market based on the Compulsory
Communications Data
• to propose transition matrices among jobs and also subgroups
transition matrices determined by some individual and job
characteristics
• to detect the main determinants of the probability of job
activation and job termination

3. Aim of this paper
• to focus on the central question whether there is sufficient
labour market dynamics in Italy
• to provide an up-to-date measure of the inflows and (partial)
outflows in the Italian labor market based on the Compulsory
Communications Data
• to propose transition matrices among jobs and also subgroups
transition matrices determined by some individual and job
characteristics
• to detect the main determinants of the probability of job
activation and job termination

4. Aim of this paper
• to focus on the central question whether there is sufficient
labour market dynamics in Italy
• to provide an up-to-date measure of the inflows and (partial)
outflows in the Italian labor market based on the Compulsory
Communications Data
• to propose transition matrices among jobs and also subgroups
transition matrices determined by some individual and job
characteristics
• to detect the main determinants of the probability of job
activation and job termination

5. Aim of this paper
• to focus on the central question whether there is sufficient
labour market dynamics in Italy
• to provide an up-to-date measure of the inflows and (partial)
outflows in the Italian labor market based on the Compulsory
Communications Data
• to propose transition matrices among jobs and also subgroups
transition matrices determined by some individual and job
characteristics
• to detect the main determinants of the probability of job
activation and job termination

6. Data
The Compulsory Communications Data (“Comunicazioni
Obbligatorie”)
• contain important information on jobs dynamics in Italy: it
includes the begin and the termination of every labour
relationship
• record all activations, transformations and (early) terminations
of employment relationships for any worker and firm since
beginning 2008
• temporary contracts started before the 1/1/2008 and ended in
the following period at the due date of termination are not
included, because firms had communicated the due date of
termination before 1/1/2008. Therefore, the labour contracts
terminations are underestimated.

7. Data
The Compulsory Communications Data (“Comunicazioni
Obbligatorie”)
• contain important information on jobs dynamics in Italy: it
includes the begin and the termination of every labour
relationship
• record all activations, transformations and (early) terminations
of employment relationships for any worker and firm since
beginning 2008
• temporary contracts started before the 1/1/2008 and ended in
the following period at the due date of termination are not
included, because firms had communicated the due date of
termination before 1/1/2008. Therefore, the labour contracts
terminations are underestimated.

8. Data
The Compulsory Communications Data (“Comunicazioni
Obbligatorie”)
• contain important information on jobs dynamics in Italy: it
includes the begin and the termination of every labour
relationship
• record all activations, transformations and (early) terminations
of employment relationships for any worker and firm since
beginning 2008
• temporary contracts started before the 1/1/2008 and ended in
the following period at the due date of termination are not
included, because firms had communicated the due date of
termination before 1/1/2008. Therefore, the labour contracts
terminations are underestimated.

9. Our dataset
• Starting from a representative sample of more than 333,000
contracts and after some data cleaning and manipulation
• we build a monthly panel dataset where the observation unit
is the job in a month, characterized by an uninterrupted (or
interrupted for a short period - 30 days) relationship between
a firm and a worker. The concept of job is different from the
one of contract. We end up with more than 263,000 jobs
• Our analyses refer to labour relationships, and not to workers:
if a worker has multiple jobs, or changes firm, or leaves a firm
and later is newly hired by the same firm (after 30 days), we
have more observations on the same worker.

10. Our dataset
• Starting from a representative sample of more than 333,000
contracts and after some data cleaning and manipulation
• we build a monthly panel dataset where the observation unit
is the job in a month, characterized by an uninterrupted (or
interrupted for a short period - 30 days) relationship between
a firm and a worker. The concept of job is different from the
one of contract. We end up with more than 263,000 jobs
• Our analyses refer to labour relationships, and not to workers:
if a worker has multiple jobs, or changes firm, or leaves a firm
and later is newly hired by the same firm (after 30 days), we
have more observations on the same worker.

11. Our dataset
• Starting from a representative sample of more than 333,000
contracts and after some data cleaning and manipulation
• we build a monthly panel dataset where the observation unit
is the job in a month, characterized by an uninterrupted (or
interrupted for a short period - 30 days) relationship between
a firm and a worker. The concept of job is different from the
one of contract. We end up with more than 263,000 jobs
• Our analyses refer to labour relationships, and not to workers:
if a worker has multiple jobs, or changes firm, or leaves a firm
and later is newly hired by the same firm (after 30 days), we
have more observations on the same worker.

12. The analysis
• Population of interest: jobs referring to all the employees that,
between March 2008 and June 2010, have been interested by
hiring, transformations, extension, stoppage of the labour
contract
• Information concerns particularly the presence of the job in
the period and the type of contract
• We build monthly transition matrices for each month between
April 2008 and June 2010
• We construct an average matrix for each year and for the
whole period of analysis and we determine the limiting vectors
• Finally, we detect the determinants of the probability of job
termination and job activation in a month

13. The analysis
• Population of interest: jobs referring to all the employees that,
between March 2008 and June 2010, have been interested by
hiring, transformations, extension, stoppage of the labour
contract
• Information concerns particularly the presence of the job in
the period and the type of contract
• We build monthly transition matrices for each month between
April 2008 and June 2010
• We construct an average matrix for each year and for the
whole period of analysis and we determine the limiting vectors
• Finally, we detect the determinants of the probability of job
termination and job activation in a month

14. The analysis
• Population of interest: jobs referring to all the employees that,
between March 2008 and June 2010, have been interested by
hiring, transformations, extension, stoppage of the labour
contract
• Information concerns particularly the presence of the job in
the period and the type of contract
• We build monthly transition matrices for each month between
April 2008 and June 2010
• We construct an average matrix for each year and for the
whole period of analysis and we determine the limiting vectors
• Finally, we detect the determinants of the probability of job
termination and job activation in a month

15. The analysis
• Population of interest: jobs referring to all the employees that,
between March 2008 and June 2010, have been interested by
hiring, transformations, extension, stoppage of the labour
contract
• Information concerns particularly the presence of the job in
the period and the type of contract
• We build monthly transition matrices for each month between
April 2008 and June 2010
• We construct an average matrix for each year and for the
whole period of analysis and we determine the limiting vectors
• Finally, we detect the determinants of the probability of job
termination and job activation in a month

16. The analysis
• Population of interest: jobs referring to all the employees that,
between March 2008 and June 2010, have been interested by
hiring, transformations, extension, stoppage of the labour
contract
• Information concerns particularly the presence of the job in
the period and the type of contract
• We build monthly transition matrices for each month between
April 2008 and June 2010
• We construct an average matrix for each year and for the
whole period of analysis and we determine the limiting vectors
• Finally, we detect the determinants of the probability of job
termination and job activation in a month

17. Average monthly transition matrices
In order to neutralize the random components in the monthly
matrices we construct an average matrix for each year and for the
whole period of analysis. The information is more stable and better
suited for a long period analysis.
Years 2008 - 2010 1 2 3 exit
1. permanent job 97.456% 0.015% 0.004% 2.525%
2. Fixed-term, Apprenticeship 1.756% 87.699% 0.030% 10.515%
3. Parasubordinate, Internship, interim 0.736% 0.671% 89.470% 9.123%
4. new entrance 30.062% 48.688% 12.292% 8.958%

18. Average monthly transition matrices, by year
Year 2008 1 2 3 exit
1. Permanent job 97.423% 0.012% 0.004% 2.562%
2. Fixed-term, Apprenticeship 1.822% 88.249% 0.022% 9.908%
3. Parasubordinate, Internship, interim 0.912% 0.738% 89.498% 8.852%
4. new entrance 29.528% 49.631% 11.699% 9.142%
Year 2009 1 2 3 Exit
1. permanent job 97.463% 0.014% 0.004% 2.518%
2. Fixed-term, Apprenticeship 1.641% 87.307% 0.026% 11.026%
3. Parasubordinate, Internship, interim 0.687% 0.608% 89.438% 9.267%
4. new entrance 30.975% 48.304% 11.933% 8.788%
Year 2010 1 2 3 Exit
1. permanent job 97.478% 0.018% 0.005% 2.499%
2. Fixed-term, Apprenticeship 1.909% 87.758% 0.049% 10.284%
3. Parasubordinate, Internship, interim 0.604% 0.704% 89.493% 9.200%
4. new entrance 28.989% 48.108% 13.864% 9.038%

19. Average monthly transition matrices, by year
Year 2008 1 2 3 Exit
1. Permanent job 97.423% 0.012% 0.004% 2.562%
2. Fixed-term, Apprenticeship 1.822% 88.249% 0.022% 9.908%
3. Parasubordinate, Internship, interim 0.912% 0.738% 89.498% 8.852%
4. new entrance 29.528% 49.631% 11.699% 9.142%
Year 2009 1 2 3 Exit
1. permanent job 97.463% 0.014% 0.004% 2.518%
2. Fixed-term, Apprenticeship 1.641% 87.307% 0.026% 11.026%
3. Parasubordinate, Internship, interim 0.687% 0.608% 89.438% 9.267%
4. new entrance 30.975% 48.304% 11.933% 8.788%
Year 2010 1 2 3 Exit
1. permanent job 97.478% 0.018% 0.005% 2.499%
2. Fixed-term, Apprenticeship 1.909% 87.758% 0.049% 10.284%
3. Parasubordinate, Internship, interim 0.604% 0.704% 89.493% 9.200%
4. new entrance 28.989% 48.108% 13.864% 9.038%

20. Average monthly transition matrices, by year
Year 2008 1 2 3 Exit
1. Permanent job 97.423% 0.012% 0.004% 2.562%
2. Fixed-term, Apprenticeship 1.822% 88.249% 0.022% 9.908%
3. Parasubordinate, Internship, interim 0.912% 0.738% 89.498% 8.852%
4. new entrance 29.528% 49.631% 11.699% 9.142%
Year 2009 1 2 3 Exit
1. permanent job 97.463% 0.014% 0.004% 2.518%
2. Fixed-term, Apprenticeship 1.641% 87.307% 0.026% 11.026%
3. Parasubordinate, Internship, interim 0.687% 0.608% 89.438% 9.267%
4. new entrance 30.975% 48.304% 11.933% 8.788%
Year 2010 1 2 3 Exit
1. permanent job 97.478% 0.018% 0.005% 2.499%
2. Fixed-term, Apprenticeship 1.909% 87.758% 0.049% 10.284%
3. Parasubordinate, Internship, interim 0.604% 0.704% 89.493% 9.200%
4. new entrance 28.989% 48.108% 13.864% 9.038%

21. Limiting vectors
Based on the average transition matrices we determine the limiting
vectors, representing the equilibrium point of a transition matrix of
a finite Markov chain. It is made up by the probabilities of
belonging to the states of the system in the long run.
Permanent job Fixed-term, Parasubordinate et al. Exit
Apprenticeship
2008 69.833% 20.194% 5.287% 4.686%
2009 71.391% 18.417% 5.442% 4.750%
2010 69.990% 18.969% 6.332% 4.708%
2008-2010 70.599% 19.082% 5.596% 4.723%

22. Subgroup analysis 2008-2010: by gender
Male 1 2 3 Exit
1. permanent job 97.171% 0.018% 0.006% 2.805%
2. Fixed-term, Apprenticeship 2.002% 87.484% 0.028% 10.486%
3. Parasubordinate, Internship, interim 0.732% 0.689% 89.731% 8.848%
4. new entrance 32.144% 47.717% 11.267% 8.872%
Limiting vector 70.563% 19.071% 5.469% 4.897%
Female 1 2 3 Exit
1. permanent job 97.806% 0.011% 0.002% 2.181%
2. Fixed-term, Apprenticeship 1.502% 87.922% 0.031% 10.545%
3. Parasubordinate, Internship, interim 0.739% 0.653% 89.209% 9.399%
4. new entrance 27.752% 49.765% 13.429% 9.054%
Limiting vector 71.171% 18.749% 5.618% 4.461%
• Probability of permanence in permanent jobs and fixed-term jobs slightly higher
for female. Transition probability from fixed-term to permanent job lower for
female.
• In the long run the proportion of permanent jobs is slightly higher for females
than for males.

23. Subgroup analysis 2080 - 2010: by geographical area
1 2 3 Exit
NE 1. permanent job 97.726% 0.011% 0.004% 2.259%
2. Fixed-term, Apprenticeship 2.533% 89.696% 0.026% 7.745%
3. Parasubordinate et al. 0.723% 0.711% 89.474% 9.092%
4. new entrance 36.004% 38.200% 16.825% 8.971%
NW 1. permanent job 97.770% 0.018% 0.005% 2.207%
2. Fixed-term, Apprenticeship 1.981% 88.120% 0.029% 9.869%
3. Parasubordinate et al. 0.694% 0.940% 88.466% 9.900%
4. new entrance 26.299% 53.216% 11.768% 8.717%
Center 1. permanent job 97.548% 0.014% 0.005% 2.433%
2. Fixed-term, Apprenticeship 1.825% 89.259% 0.040% 8.876%
3. Parasubordinate et al. 0.639% 0.667% 90.267% 8.427%
4. new entrance 30.824% 44.990% 13.565% 10.621%
South 1. permanent job 96.761% 0.016% 0.003% 3.220%
2. Fixed-term, Apprenticeship 0.929% 84.425% 0.028% 14.618%
3. Parasubordinate et al. 0.870% 0.408% 89.713% 9.009%
4. new entrance 27.442% 55.965% 8.595% 7.998%
Islands 1. permanent job 96.799% 0.018% 0.003% 3.180%
2. Fixed-term, Apprenticeship 1.002% 85.569% 0.024% 13.405%
3. Parasubordinate et al. 0.914% 0.378% 89.051% 9.657%
4. new entrance 28.904% 53.563% 9.120% 8.412%

24. Limiting vectors: by geographical area
Permanent job Fixed-term, Parasubordinate et al. Exit
Apprenticeship
NE 76.156% 14.184% 5.967% 3.693%
NW 70.963% 20.099% 4.555% 4.382%
Center 70.612% 18.800% 6.213% 4.375%
South 64.025% 23.828% 5.574% 6.572%
Islands 65.132% 23.369% 5.263% 6.236%
• In the northern regions: Higher probabilities of permanence, higher probability
of transition from temporary job to permanent job.
• In the long run, higher proportion of permanent jobs and lower proportion of
temporary jobs in the North than in the South.

25. Subgroup analysis: by age of the worker
35 and younger 1 2 3 Exit
1. permanent job 97.435% 0.020% 0.003% 2.542%
2. Fixed-term, Apprenticeship 1.853% 88.576% 0.030% 9.541%
3. Parasubordinate, Internship, interim 0.868% 0.933% 87.958% 10.241%
4. new entrance 26.703% 48.867% 14.844% 9.585%
36-50 years old 1 2 3 Exit
1. permanent job 97.751% 0.011% 0.002% 2.236%
2. Fixed-term, Apprenticeship 1.777% 86.834% 0.028% 11.360%
3. Parasubordinate, Internship, interim 0.666% 0.457% 90.436% 8.441%
4. new entrance 35.266% 48.002% 8.598% 8.133%
51 and older 1 2 3 Exit
1. permanent job 96.895% 0.011% 0.011% 3.083%
2. Fixed-term, Apprenticeship 1.193% 85.267% 0.035% 13.505%
3. Parasubordinate, Internship, interim 0.411% 0.134% 92.946% 6.509%
4. new entrance 31.406% 49.576% 10.656% 8.363%

26. Limiting vectors: by age of the worker
Permanent job Fixed-term, Apprentiship et al. Exit
Apprenticeship
35 and younger 67.817% 21.308% 6.036% 4.839%
36-50 years old 77.093% 15.086% 3.735% 4.085%
51 and older 66.199% 19.301% 8.803% 5.697%
Note that in our dataset on average 67% of the jobs are for 35 and younger, 25% for
36-50 years old and only 7% for 51 and older.

27. Subgroup analysis: by citizenship of the worker
Not Italian 1 2 3 Exit
1. permanent job 98.399% 0.012% 0.001% 1.587%
2. Fixed-term, Apprenticeship 1.986% 85.578% 0.017% 12.420%
3. Parasubordinate, Internship, interim 0.660% 0.943% 86.475% 11.923%
4. new entrance 46.686% 41.473% 5.281% 6.561%
Limiting vector 88.398% 7.855% 1.065% 2.681%
Italian 1 2 3 Exit
1. permanent job 97.084% 0.016% 0.005% 2.895%
2. Fixed-term, Apprenticeship 1.706% 88.161% 0.033% 10.100%
3. Parasubordinate, Internship, interim 0.743% 0.645% 89.759% 8.853%
4. new entrance 24.476% 51.113% 14.648% 9.764%
Limiting vector 62.308% 24.232% 7.968% 5.493%
In the long run, the most common type of job signed by the immigrants is the
permanent job.

28. Subgroup analysis: by education of the worker
N.A. 1 2 3 Exit
1. permanent job 98.181% 0.012% 0.002% 1.805%
2. Fixed-term, Apprenticeship 1.761% 85.283% 0.029% 12.927%
3. Parasubordinate, Internship, interim 0.589% 0.638% 88.903% 9.870%
4. new entrance 45.327% 40.287% 6.928% 7.458%
Compulsory edu 1 2 3 Exit
1. permanent job 96.739% 0.018% 0.005% 3.238%
2. Fixed-term, Apprenticeship 1.893% 86.116% 0.021% 11.970%
3. Parasubordinate, Internship, interim 0.823% 0.675% 88.517% 9.984%
4. new entrance 26.998% 56.089% 8.500% 8.413%
Secondary edu 1 2 3 Exit
1. permanent job 97.527% 0.014% 0.006% 2.453%
2. Fixed-term, Apprenticeship 1.753% 89.695% 0.041% 8.511%
3. Parasubordinate, Internship, interim 0.748% 0.763% 89.413% 9.075%
4. new entrance 23.638% 46.692% 18.140% 11.530%
College 1 2 3 Exit
1. permanent job 98.154% 0.008% 0.007% 1.832%
2. Fixed-term, Apprenticeship 1.268% 91.743% 0.033% 6.956%
3. Parasubordinate, Internship, interim 0.697% 0.538% 91.042% 7.723%
4. new entrance 20.912% 42.825% 27.979% 8.283%

29. Limiting vectors: by education of the worker
Permanent job Fixed-term, Parasubordinate et al. Exit
Apprenticeship
N.A. 86.277% 8.648% 1.974% 3.102%
Compulsory edu 64.985% 24.514% 4.508% 5.993%
Secondary edu 64.372% 22.472% 8.353% 4.803%
College 62.925% 20.966% 12.232% 3.877%
Note that more than 60% of N.A. are immigrants.

30. Inflows and outflows determinants
Based on the information on the presence/absence of each job in
the each month, we estimate
• the probability of job termination (that the job is present at
time t − 1 and not present at time t)
• the probability of job activation (that the job is not present at
time t − 1 and is present at time t)
along with some individual and job characteristics.

31. Logit estimates for job inflows and outflows, β
coefficients
in out
contract: fixed-term, apprenticeship -0.194 *** 1.219 ***
contract: Parasubordinate et al. -0.105 *** 1.370 ***
Italian -0.135 *** 0.139 ***
North-East -0.041 *** -0.483 ***
North-West -0.006 -0.364 ***
Center -0.016 -0.391 ***
South -0.031 *** 0.022 **
edu: N.A. 0.061 *** -0.064 ***
edu: secondary 0.039 *** -0.182 ***
edu: college 0.145 *** -0.274 ***
female 0.044 *** 0.070 ***
age2008 0.002 *** -0.006 ***
N 4,484,400 2,281,652
Controls for year, occupations, sectors.
Note: reference categories are: male, with compulsory education, with no
Italian citizenship, with permanent employment, in the Islands.

32. Comments to the logistic regression
• the probability of job activation (in): higher for the older,
higher for female, higher for secondary education and college,
lower if jobs are signed by Italians;
• the probability of job termination (out): lower for the older,
higher for female, lower for secondary education and college,
higher if jobs are signed by Italians, lower in the North and in
the Center, higher for temporary jobs.

33. Concluding remarks
• Proposal of transition matrices related to jobs rather than
workers
• The potentiality of the CC data: working on the population
rather than a sample, creating even daily transition matrices
• Reconstructing the worker’s history

34. Concluding remarks
• Proposal of transition matrices related to jobs rather than
workers
• The potentiality of the CC data: working on the population
rather than a sample, creating even daily transition matrices
• Reconstructing the worker’s history

35. Concluding remarks
• Proposal of transition matrices related to jobs rather than
workers
• The potentiality of the CC data: working on the population
rather than a sample, creating even daily transition matrices
• Reconstructing the worker’s history