EXPLAINING THE SPREAD OF TEMPORARY JOBS AND ITS IMPACT ON LABOR TURNOVER

• Author: Olivier Charlot,Pierre Cahuc,Franck Malherbet
• DOI: http://doi.org/10.1111/iere.12167
• Published date: 01 May 2016
• Date: 01 May 2016

INTERNATIONAL ECONOMIC REVIEW

Vol. 57, No. 2, May 2016

EXPLAINING THE SPREAD OF TEMPORARY JOBS AND ITS IMPACT ON LABOR

TURNOVER∗

BYPIERRE CAHUC,OLIVIER CHARLOT,AND FRANCK MALHERBET 1

CREST,Ecole Polytechnique,France, and IZA, CEPR; University of Cergy-Pontoise, THEMA,

France; CREST, University of Rouen, France, and IZA

This article provides a simple model that explains the choice between permanent and temporary jobs. This model,

which incorporates important features of actual employment protection legislations neglected by the economic literature

so far, reproduces the main stylized facts about entries into permanent and temporary jobs observed in Continental

Europeancountries. We find that job protection has very small effects on total employment but induceslarge substitution

of temporary jobs for permanent jobs, which significantly reduces aggregate production.

1. INTRODUCTION

It is recurrently argued that the dramatic spread of temporary jobs in Continental European

countries is the consequence of the combination of stringent legal constraints on the termination

of permanent jobs and of weak constraints on the creation of temporary jobs.2Strikingly,

however, very little is known about the creation of temporary and permanent jobs, inasmuch

as very few contributions have analyzed the choice between these two types of job. There are

also very few explanations of the duration of temporary jobs.

Our article contributes to filling this gap. It provides a model that explains the duration of

temporary jobs and the choice between temporary and permanent jobs. This model reproduces

important stylized facts that previous models were unable to explain. In particular, for countries

with stringent job protection, the model fits the large share of temporary contracts in employ-

ment inflows, the huge amount of creation of temporary contracts of very short duration, and

the large contribution of inflows into temporary jobs to fluctuations in employment inflows

overall. The model sheds new light on the consequences of employment protection. It shows

that the stringency of legal constraints on the termination of permanent jobs has very little

effect on total employment, but does induce a large-scale substitution of temporary jobs for

permanent jobs, which significantly reduces aggregate production.

One of the main originalities of our approach is to account for important features of employ-

ment protection legislations that have been neglected by the literature so far. In most countries,

it is costly to dismiss temporary workers before the date of termination of the contract stipulated

when the job starts. More precisely, in the “French type” regulation, which prevails in Belgium,

France, Greece, Italy, and Germany, temporary contracts cannot be terminated before their

∗Manuscript received August 2013; revised December 2014.

1We thank the editor (Guido Menzio) and three anonymous referees for helpful comments and suggestions. We also

thank Samuel Bentolila, Tito Boeri, Werner Eichhorst, Øivind Nilsen, Oskar Nordstr¨

om Skans, Jos´

e Ignacio Garc´

ıa

P´

erez, Pedro Portugal, Kostas Tatsiramos, Bruno Van der Linden, and Frank Walsh for providing us with information

about employment protection legislation and temporary jobs in different OECD countries. We thank Fabio Berton,

Herv´

e Boulhol, Juan Dolado, Bruno Decreuse, Renato Faccini, Pietro Garibaldi, Juan Jimeno, Etienne Lehmann,

Jean-Baptiste Michau, Claudio Michelacci, Fabien Postel-Vinay, Francesco Pappad`

a, Barbara Petrongolo, Jean-Luc

Prigent, Jose Silva, Eric Smith, and H´

el`

ene Turon for useful comments. We thank the Chaire S´

ecurisation des parcours

professionnels and the Labex Ecodec for their support. Please address correspondence to: Pierre Cahuc, CREST-LMA,

15 Boulevard Gabriel P´

eri, 92245 Malakoff Cedex, France. Phone: +33-1411-737-17. Fax: +33-1411-760-29. E-mail:

pierre.cahuc@ensae.fr.

2See Boeri (2011) for a synthesis.

533

C

(2016) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

534 CAHUC,CHARLOT,AND MALHERBET

expiration date,3whereas in the “Spanish type” regulation, which covers Spain and Portugal,

the rule for dismissals before the expiration date of temporary contracts is the same as for

permanent contracts.4Hence, for a given employment spell, it is generally at least as costly to

terminate a temporary contract before its expiration date as it is to terminate a regular contract.

In the previous literature, it is generally assumed that it is costly to terminate permanent con-

tracts, whereas temporary contracts can be terminated at no cost at any time. This assumption,

made for the sake of technical simplicity, is at odds with many actual regulations. It implies

that employers prefer temporary jobs, which can be destroyed at no cost, to permanent jobs,

which are costly to destroy, thus making it difficult to explain the choice between permanent

and temporary jobs. Our more realistic approach assumes that temporary contracts cannot be

terminated before their expiration date.

We consider a job search and matching model where firms hire workers to exploit production

opportunities of different expected durations. Some production opportunities are expected to

end (i.e., to become unproductive) quickly; others are expected to last longer. This assumption

takes into account the heterogeneity of expected durations of production opportunities, which is

an important feature of modern economies. For instance, firms can get orders for their products

for periods of several days, several months, or several years, and it is not certain that these

orders will be renewed. In the model, jobs can be either permanent or temporary. Permanent

employees are protected by dismissal costs. Temporary jobs can be destroyed at zero cost at

their expiration date, which is chosen at the instant when workers are hired. But employers

have to keep and pay their employees until the date of termination of temporary jobs. These

assumptions about employment legislation, which are framed to match the main features of

Continental European labor regulations, do not induce Pareto-optimal allocations. However,

permanent workers protected by firing costs may give moral and political support to such

regulations.5

When firing costs are sufficiently small, we find that all production opportunities are exploited

with permanent jobs. When firing costs are relatively large, permanent jobs are chosen to exploit

production opportunities expected to endure for a long time, whereas temporary jobs are used

for production opportunities with short expected durations. In this framework, higher firing

costs increase the share of entries into temporary jobs.

We show that our model matches the main stylized facts concerning entries into permanent

and temporary jobs in Continental European countries. Moreover, simulation exercises show

that the durations of temporary jobs are much shorter than the durations of production op-

portunities. Therefore, higher firing costs, by increasing the share of temporary jobs, induce

a strong excess of labor turnover on production opportunities with relatively short durations.

This excess of labor turnover is detrimental to temporary workers whose expected job duration

becomes shorter when the employment protection of permanent jobs becomes more stringent.

In this context, heightened protection for permanent jobs will have very small negative effects

on aggregate employment. However, this small aggregate impact is the net consequence of two

large counteracting effects: a strong decrease in the number of permanent jobs and a strong

increase in the number of temporary jobs. This large reallocation of jobs, which conforms

to empirical evidence,6decreases aggregate production, because the production (net of labor

turnover costs) of temporary jobs is much smaller than that of permanent jobs. All in all, our

model shows that protection of permanent jobs has very small effects on aggregate employment

3There are obviously exceptions to this general rule, for instance, for misbehavior on the part of one of the parties.

The legislations are described in Section A.1 of the Appendix. For a given employment spell, it appears that it is

generally at least as costly to terminate a temporary contract before its date of termination as to terminate a regular

contract.

4Henceforth, we focus on regulation of the French type. We show in Cahuc et al. (2012) that the Spanish type yields

the same outcome as the French type in the context of our model.

5See Saint-Paul (1996, 2002).

6See, among others, Autor (2003), Kahn (2010), Centeno and Novo (2011), Cappellari et al. (2012), Hijzen et al.

(2013).

TEMPORARY JOBS AND LABOR TURNOVER 535

but induces employment composition effects that significantly reduce aggregate production.

Changes in aggregate production are five times larger than changes in aggregate employment.

Our article is related to at least three strands of the literature.

First, we introduce heterogeneity of idiosyncratic productivity shock arrival rates into the

job search model. This allows us to explain empirical evidence that indicates that the expected

duration of production opportunities is a major motive for using temporary jobs when the de-

struction of permanent jobs is costly. Indeed, it turns out that the share of temporary contracts

is higher in industries with higher labor turnover in countries with stringent job protection

(Bassanini and Garnero, 2013). Drager and Marx (2012) find, using a large firm-level data

set from 20 countries, that workload fluctuations strongly increase the probability of hiring

temporary workers in rigid labor markets but that no such effect is observed in flexible la-

bor markets. Strikingly, we are not aware of any model that explains such facts. Our model

sheds light on the impact of temporary contracts from a perspective different from the one in

which temporary contracts are viewed as a way of screening workers before they are promoted

into permanent jobs.7Actually, in all countries, permanent contracts comprise probationary

periods, with no firing cost and very short notice, which are used to screen workers into per-

manent jobs. The maximum mandatory duration of probationary periods is around several

months, depending on countries, industries, and skills.8To the extent that temporary jobs can-

not be terminated before their expiration date, it can only be profitable to screen workers by

means of temporary contracts if the duration of the probationary period is too short, at least

shorter than that of temporary contracts.9Accordingly, the view that temporary contracts are

used to screen workers can be useful to explain the spread of temporary jobs lasting longer

than the probationary period of permanent jobs. But this approach cannot explain the huge

amount of creation of temporary contracts of very short spell, much shorter than that of pro-

bationary periods.10 For instance, in France, the average duration of temporary jobs is about

one month and a half, whereas the probationary periods last at least 2 months and can go to

8 months.11

Second, we complement the literature on the impact of employment protection legislation12

by explaining the choice between permanent and temporary jobs. Most of this literature does

not explain this choice.13 Usually, in this literature, temporary jobs, which can be destroyed at

zero cost, are preferred to permanent jobs, which are costly to destroy, and it is either assumed

that all new jobs are temporary or that the regulation forces firms to create permanent jobs.

As far as we know, four papers explain the choice between temporary and permanent jobs

in a dynamic setting.14 Berton and Garibaldi (2012) propose a matching model with directed

7See Bucher (2010), Faccini (2014), Kahn (2010), Nagypal (2002), Portugal and Varej˜

ao (2009).

8See: http://www.ilo.org/dyn/eplex/termmain.home?p_lang=uk.

9In general, the probationary period of temporary jobs is much shorter than that of permanent jobs. Furthermore,

when a temporary job is transformed into a permanent job, the duration of the temporary job has to be subtracted from

the duration of the probationary period of the permanent job.

10 To the extent that workers can be dismissed at zero cost during probationary periods, at first sight it is more

profitable to exploit job opportunities expected not to last long with permanent contracts that are terminated at no

cost during the probationary periods, instead of with temporary contracts that cannot be terminated before their date

of termination even if the job becomes nonprofitable. However, this type of behavior is illegal. An employer who

systematically hires workers under permanent contracts and dismisses them during the probationary period instead of

using temporary contracts runs the risk of being prosecuted. Our article does not account for probationary periods,

which are left for future research. We merely assume that permanent workers are protected by firing costs from the

start of their contract.

11 In France, the legal maximum duration of the probationary period for permanent contract goes from 2 months

for blue collar workers to 4 months for white collar workers. The probationary period can be renewed once if this is

stipulated in the labor contract.

12 See, among others, Lazear (1990), Bentolila and Saint-Paul (1992), Saint-Paul (1996), Ljungqvist (2002), and

l’Haridon and Malherbet (2009).

13 See, among others, Blanchard and Landier (2002), Cahuc and Postel-Vinay (2002), Boeri and Garibaldi (2007),

Sala, Silva and Toledo (2012), Costain et al. (2010), Bentolila et al. (2012), and Saint-Paul (1996).

14 Kahn (2010) provides a static two period model where temporary jobs are used to screen workers.