The role of temporary contracts and unemployment benefits

for worker careers: EVIDENCE FROM SPAIN

J. IGNACIO GARCIA PEREZUNIVERSIDAD PABLO DE OLAVIDE & FEDEA

Joint EC/OECD Seminar “Job Quality, LabourMarket Performance and Well-Being”

Brussels, 27-28 November 2014

INTRODUCTION

FIXED-TERM CONTRACTS (lower firing costs)

one of the most popular measures to fight against youth unemployment in Europe

Do fixed -term contracts really help?

The theory:

Impact FTC on labor market outcomes of affected workers Ambiguous!!

Do fixed -term contracts really help?

• Empirical Evidence:

• Potential interactions with Unemployment Benefits: incidence on individuals’ labour market paths

TC(3.366)

TC783

(23.3%)

PC325

(9.7%)

U1890

(56.1%)

TC235

(30.0%)

PC144

(18.4%)

U302

(38.6%)

C102

(13.0%)

C368

(10.9%)

TC1526

(80.7%)

PC206

(10.9%)

C158

(8.4%)

TC82

(34.9%)

PC42

(17.9%)

U79

(33.6%)

C32

(13.6%)

TC245

(81.1%)

PC36

(11.9%)

C21

(6.9%)

TC356

(23.3%)

PC93

(6.1%)

U932

(61.1%)

C145

(9.5%)

Only 25% Only 25% skilled skilled

workers workers have a PC have a PC after these after these transitions transitions (less than (less than

18% among 18% among the the

unskilled)unskilled)

Transitions into permanent employment in Spain: An Empirical analysis for young workers (García-Perez & Muñoz-Bullón, BJIR-2009)

Transitions into permanent employment in Spain: An Empirical analysis for young workers (García-Perez & Muñoz-Bullón, BJIR-2009)

� Multiple-spell duration techniques applied to a longitudinal data set of temporary workers for the period 1996-2003.

� MAIN RESULTS:1. Transitions into permanent employment do not increase with

tenure.

2. Indeed, as tenure in the temporary contract lengthens, the entryinto another temporary contract rises.

3. But the transition rate into permanent work is not higher than the transition rate into unemployment.

4. Furthermore, individuals with high duration of unemployment flow into permanent work less frequently.

The dual labour market TRAP: PERSISTENCE(Conde-Ruiz, Felgueroso & García Pérez, 2011)

% with a temporary contract by entry age (MCVL-2009, 1990-2009)

SPAIN

0

10

20

30

40

50

60

70

80

90

100

15 20 25 30 35 40 45 50 55 60 65

Age

begin working at 16-20

begin working at 21-25

begin working at 26-30

begin working at 31-35

• Long-term impact of FTC on affected workers’ careers is an open question.• Furthermore, the incidence of Unemployment Benefits on individuals’ labour market paths is also not clear.

Do fixed -term contracts really help?

Are Unemployment Benefits harmful to the stability of working careers? The case of Spain

Forthcoming in SERIEs – Journal of the Spanish Economic Association

YOLANDA F. REBOLLO-SANZ J. IGNACIO GARCÍA-PEREZU. PABLO DE OLAVIDE U. PABLO DE OLAVIDE & FEDEA

What do we do?

� The goal of this paper is to assess the incidence of the Unemployment Insurance System (UIS) on individuals’labour market paths.

� In particular, we analyze the incidence of the entitlement length of benefits using a sample of Spanish male workers during the period 1995-2007.

� This medium-term perspective is important, specially in dual labour markets: the temporality trap.

� By using a “Timing of Event Approach”, we estimate a Multivariate Duration Model.

What we find?

� There is a strong relationship between UB entitlement length and unemployment duration

� But at the same time, there is a positive but weak relationship between entitlement length and employment duration

�UB might encourage job stability for temporary workers by increasing :

�Subsequent job tenure: the pure matching effect

�The entrance probability into a permanent contract from a temporary contract

The Estimated Unemployment Exit Probability in relation to the UB entitlement length

12

0.1

.2.3

.4E

xit P

roba

bilit

y

0 2 4 6 8 10Months of Unemployment

No Benefits PBD=6

PBD=8

Previous contract:TC

0.1

.2.3

.4E

xit P

roba

bilit

y

0 2 4 6 8 10Months of Unemployment

No Benefits PBD=6

PBD=8

Previous contract:PC

PBD=lentgh of the benefit entitlement

Measuring the UIB exhaustion effect

Estimated Employment Exit Probability in relation to UB entitlement

Layoff Quit JJ-PC JJ-TC Empl.

Duration

Contract Type: TC

Non-insured 7.03% 1.85% 1.64% 3.98% 5.96

UIS entitlement:

4 6.20% 1.36% 1.73% 3.88% 6.59

6 5.98% 1.24% 1.76% 3.85% 6.77

8 5.83% 1.16% 1.78% 3.83% 6.90

12 5.62% 1.06% 1.81% 3.81% 7.08

24 5.27% 0.91% 1.86% 3.76% 7.39

Contract Type: PC

Non-insured 0.44% 0.36% 0.62% 0.37% 42.52

UIS entitlement:

4 0.45% 0.29% 0.58% 0.47% 43.25

6 0.46% 0.28% 0.58% 0.46% 43.44

8 0.46% 0.25% 0.58% 0.45% 43.57

12 0.46% 0.24% 0.58% 0.44% 43.73

24 0.47% 0.21% 0.58% 0.43% 44.00

MAIN RESULT

� We find a low impact of changes in UB entitlements over workers careers:

� By reducing the entitlement length:� Workers are employed only a bit more…

� … but at the expense of suffering more transitions from employment to unemployment.

� Thus, shortening the duration of the benefit entitlement period (by 50%) leads to small gains in overall employment stability (increases only by 4.3%).

CAN FIXED-TERM CONTRACTS PUT LOW SKILLED YOUTH ON A

BETTER CAREER PATH? EVIDENCE FROM SPAIN

J. IGNACIO GARCIA PEREZUNIVERSIDAD PABLO DE OLAVIDE & FEDEA

IOANA MARINESCUHARRIS SCHOOL OF PUBLIC POLICY – UNIVERSITY OF CHICAGO

JUDIT VALL CASTELLOUNIVERSITAT DE GIRONA & CENTRE FOR RESEARCH IN

ECONOMICS AND HEALTH, UNIVERSITAT POMPEU FABRA

What do we do?

� In this paper we use a regression discontinuity regression discontinuity design design that exploits a large change in Spanish regulation.

� Spain is an ideal ground for our research because fixed-term contract use was liberalized in 1984

� We innovate by examining the longlong --term impact of term impact of fixedfixed --term contractsterm contracts on young worker’s career by using Social Security data (more than 20 years of follow up).

Empirical Approach

Empirical ApproachReform in 1984:

13,349 8,327

� We will use two complementary methodologies:

1. Moulton (2011): cohort regression discontinuity design applied to long-term outcomes for workers career.

�� Accumulated impact of the reformAccumulated impact of the reform (since LM entry up until 2006).

2. Oreopoulos et al. (2012): collapse the individual-level data by birth cohort, calendar year and years of potential experience

�� Yearly effect of the reformYearly effect of the reform measured for an average year in a worker 's career

Empirical Approach

First Empirical Approach (Moulton, 2011)

ijicii

ciici

tryUnemRateEnSectorFEreformCBirthYearCBirthYear

reformCBirthYearCBirthYearreformOutcome

εββββββα

+++−+−

+−+−++=

62

52

4

321

)()(

)()(

With )( CBirthYearreform ic ≥Ι=

Results: Number of days worked until 2006

effect -205.21*** -201.30*** -348.08*** -306.54*** -339.34*** -315.26***(69.23) (59.67) (65.63) (56.37) (57.55) (54.37)

trend -173.37*** -165.72*** -125.01*** -124.58*** -124.99*** -124.60***(29.89) (27.12) (26.91) (25.99) (26.90) (25.99)

posttrend -179.35*** -183.50*** -222.68*** -230.49*** -228.17*** -225.02***(44.08) (37.83) (54.23) (44.36) (42.45) (37.76)

trend2 6.72*** 7.43*** 10.54*** 10.68*** 10.54*** 10.68***(2.05) (1.87) (1.70) (1.66) (1.70) (1.66)

posttrend2 10.10** 9.33** 5.63 6.84 6.34 6.13*(4.55) (3.47) (6.25) (4.24) (4.43) (3.33)

ur -42.12*** -43.13*** -42.40*** -43.39*** -42.40*** -43.38***(7.46) (6.97) (7.53) (7.04) (7.53) (7.03)

Sector X X XConstant 6,040.96*** 4,617.74*** 6,181.09*** 4,735.86*** 6,181.27*** 4,735.73***

(204.25) (188.00) (209.45) (180.52) (209.32) (180.43)

Observations 21,676 21,676 21,676 21,676 21,676 21,676R-squared 0.38 0.43 0.38 0.43 0.38 0.43

*** p<0.01, ** p<0.05, * p<0.1

ALL COHORTS EXCLUDING 68&69 COHORTS EXCLUDING 1968 COHORT

Robust standard errors in parentheses

Results: Log of Accumulated Wages until 2006

effect -0.20*** -0.18** -0.35*** -0.34*** -0.37*** -0.36***(0.07) (0.07) (0.08) (0.06) (0.07) (0.06)

trend 0.05 0.04 0.11*** 0.10*** 0.11*** 0.10***(0.03) (0.03) (0.03) (0.02) (0.03) (0.02)

posttrend -0.16*** -0.15*** -0.24*** -0.23*** -0.23*** -0.22***(0.03) (0.03) (0.03) (0.03) (0.03) (0.03)

trend2 0.00* 0.00* 0.01*** 0.01*** 0.01*** 0.01***(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

posttrend2 0.01** 0.01** 0.00 0.00 0.00 0.00(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

ur -0.02*** -0.02*** -0.02*** -0.02*** -0.02*** -0.02***(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

Sector X X XConstant 12.32*** 12.16*** 12.51*** 12.34*** 12.50*** 12.34***

(0.08) (0.11) (0.09) (0.12) (0.09) (0.12)

Observations 14,793 14,747 14,793 14,747 14,793 14,747R-squared 0.04 0.08 0.05 0.08 0.05 0.08

*** p<0.01, ** p<0.05, * p<0.1

ALL COHORTS EXCLUDING 68&69 COHORTS EXCLUDING 1968 COHORT

Robust standard errors in parentheses

� The previous approach does not properly takes into account changes in the labour market through the analyzed period.

� We now collapse the data by birth cohort (c), calendar year (t) and years of potential experience (e).

� We then run a basic specification (controlling for cell sizes):

� θc is the cohort fixed effect;

� ϕt is a calendar year fixed effect.

� γe is a potential experience fixed effect.

ctetcct uy ++++= γφθα

Second Empirical Approach (Oreopoulos et al. 2012)

The effect on the number of daysworked per year is negative :

Results are almost the same with a linear trend and using quarter instead of year of birth)

Experience Experience + Excluding + Excluding interactionsNo FE Time FE FE & Time FE 68&69 cohorts 1968 cohort with Experience

effect -8.670*** -5.238** -9.841*** -5.971** -9.443* -8.779** -3.569(2.590) (2.065) (2.916) (2.350) (4.772) (3.722) (5.328)

reformrex1 -15.772**(6.730)

reformrex2 -22.489**(8.662)

reformrex3 -31.751***(7.594)

reformrex4 -42.019***(7.031)

reformrex5 -18.738**(6.403)

Constant 257.469*** 65.701*** 334.642*** 188.594*** 195.963*** 196.006*** 191.591***(2.499) (1.827) (2.969) (12.989) (13.990) (13.948) (13.660)

Observations 2,080 2,080 2,080 2,080 2,080 2,080 2,080R-squared 0.033 0.686 0.712 0.822 0.822 0.822 0.831Robust standard errors in parentheses*** p<0.01, ** p<0.05, * p<0.1

The effect of the reform on the number of days worked is negative during the first 5 years of real experience

-60

-40

-20

020

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20years of real experience

The effect on wages is alsonegative :

Experience Experience + Excluding + Excluding interactionsNo FE Time FE FE & Time FE 68&69 cohorts 1968 cohort with Experience

effect -0.153*** -0.109*** -0.171*** -0.130*** -0.201*** -0.220*** -0.125***(0.041) (0.031) (0.047) (0.040) (0.021) (0.018) (0.030)

reformrex1 -0.318***(0.051)

reformrex2 -0.167**(0.075)

reformrex3 -0.158**(0.068)

reformrex4 -0.251***(0.055)

reformrex5 -0.147**(0.050)

reformrex6 -0.158***(0.044)

reformrex7 -0.138***(0.039)

reformrex8 -0.090**(0.032)

reformrex9 -0.043*(0.024)

Constant 9.196*** 7.818*** 9.980*** 9.641*** 9.740*** 9.739*** 9.711***(0.040) (0.069) (0.056) (0.081) (0.075) (0.075) (0.076)

Observations 1,457 1,457 1,457 1,457 1,457 1,457 1,457R-squared 0.027 0.831 0.860 0.906 0.907 0.907 0.911Robust standard errors in parentheses*** p<0.01, ** p<0.05, * p<0.1

The effect of the reform on yearly wages is still negative after 9 years of real experience

-.4

-.3

-.2

-.1

0.1

1 2 3 4 5 6 7 8 9 10

11

12

13

14

15

16

17

18

19

20

years of real experience

Concluding remarks: FINDINGS

� Widely available fixed-term contracts at labor market entry means:

� More likely to be on a fixed-term contract in the long run

� Higher use of fixed-term contracts during the LM career

� Fewer days worked (-315)

� Lower wages (-36%)

� We also find an increase in the probability of finding a job before age 19

Concluding remarks: FINDINGS

� When controlling for what happens in the labor market during the 20-year span we analyze

� We get similar results:� (-8.78) days of work per year (193 less accum. days)

� (-22%) less yearly wages

� We prefer these set of results because they are more carefully taking into account age of LM entry and business cycle effects.

Concluding remarks: POLICY

� Making FTC more readily available reduced the welfare of low skilled workers .

� FTC allow low skilled workers to get a quicker entry into the LM but the long-run consequences are negative .

� We conclude that, far from being a stepping stone, fixedfixed --term contracts are a stumbling block for the term contracts are a stumbling block for the career of low skilled workerscareer of low skilled workers (vicious circle of instability).

� We can adapt our RD approach to this specification by estimating:

� Reformc is, again, a dummy equal to 1 for cohorts 1969 and later

� ϕt is a calendar year fixed effect. � γe is a potential experience fixed effect.

Second Empirical Approach (Oreopoulos et al. 2012)

ctetccc

ccccct

ureformCBirthYearCBirthYear

reformCBirthYearCBirthYearreformy

+++−+−+

−+−++=

γφβββββα

25

24

321

)()(

)()(

ROBUSTNESS CHECK:

Sample restricted to include only those who began working at ages 14-17 or to

include in the control group only those who begin working before the reform

Days worked

-296.86*** -212.67** -384.48*** -257.24* -452.30*** -347.97**

(100.53) (95.21) (125.87) (131.70) (116.17) (129.88)

-921.75*** -887.81*** -954.67*** -897.57*** -946.40*** -905.91***

(77.60) (75.26) (107.41) (104.91) (99.69) (100.32)

Control Group: First Year Employment Before Reform

ALL COHORTS EXCL. 68&69 COHORTS EXCL. 1968 COHORT

Age First Job 14-17

Accumulated wages (logs)

Age First Job 14-17 -0.22** -0.21** -0.50*** -0.44*** -0.57*** -0.54***(0.08) (0.08) (0.06) (0.05) (0.03) (0.04)

Control Group: First Year -0.43*** -0.40*** -0.49*** -0.47*** -0.51*** -0.49***Employment Before Reform (0.07) (0.08) (0.05) (0.05) (0.04) (0.05)

ALL COHORTS EXCL. 68&69 COHORTS EXCL. 1968 COHORT