Abstract
Firing costs are often blamed for unemployment. This paper investigates this
well spread belief. The main points are two. First, firing costs are modelled in
an efficiency wage model to capture their effects on employment through wages.
Secondly, dismissal conflicts are modelled explicitly. In the context of
imperfectly observable effort, a double moral hazard problem can arise and in
turn firing costs reduced employment because they increase the rent to be paid
to workers. The determinants of the double moral hazard problem such as the
imprecise definition of dismissal causes are analysed. The main policy
conclusion is that focus should move onto the clarification of the different
causes of dismissal to minimise the room of interpretation. If so, then high
enough severance payments in case of “unfair” dismissals can actually have a
punishment role and prevent the double moral hazard problem.
This paper was produced as part of the Centre’s
Labour Markets Programme
Employment Protection and Unemployment
in an Efficiency Wage Model
Maia Güell
July 2000
Series Editor: Graham Ingham
Published by
Centre for Economic Performance
London School of Economics and Political Science
Houghton Street
London WC2A 2AE
 Maia Güell, submitted October 1999
ISBN 0 7530 1389 4
Individual copy price: £5
Employment Protection and Unemployment in an
Efficiency Wage Model
Maia Güell
1.
Introduction
2.
The Model
2.1
Redundancies and disciplinary dismissals in conflict
2.2
Non-shirking condition
2.3
Hiring decisions
2.4
Market equilibrium
2.5
Policy implications
4
4
6
8
8
11
3.
Conclusion
12
References
1
14
The Centre for Economic Performance is financed by the Economic and Social Research Council
Acknowledgements
I would like to thank Alan Manning and Gilles Saint-Paul and very especially Miguel Angel
Malo for useful discussions.
Maia Ghell is a member of the Centre for Economic Performance and the Industrial Relations
Department, Princeton University.
EmploymentP rotectionandU nemployment
inanE¢ciencyW ageM odel
M aiaG üell
1
Introduction
FiringcostsareoftenblamedforunemploymentinEurope(seeO ECD (19 9 5), forinstance).
T he aim ofthis paperis toinvestigate this widespread belieffrom a theoreticalpointof
view. T he modelwe build makes two main points. First, …ring costs are introduced in
an e¢ciency wage modeltocapture theire¤ ects on employmentthrough wages. Second,
dismissalcon‡icts aremodeledexplicitlyandtheircostis derived. Inparticular, twotypes
ofdismissalsareconsidered, redundanciesanddisciplinarydismissals, whereemployersand
employees havecon‡ictinginterests1 .
D espite the prevalentideaofthe (negative) e¤ ectof…ringcosts on employment, speciallyamongpolicymakersandemployers, thereareverydi¤ erentviewsamongeconomists
dependingon the modelused. Forinstance, accordingtothe insider-outsidertheory put
forward by L indbeck and Snower (19 8 8), …ring costs are a source ofmarket powerfor
incumbentworkers (the insiders) vis-a-vis the unemployed (the outsiders). Insiders use
theirmarketpowerto exercise upward pressure on theirwages and thereby generating
unemployment. A ccordingtothis view, the higherthe …ringcosts, the higherthe unemployment.
A completelydi¤ erentviewof…ringcosts is theoneby L azear(19 9 0). H eshows that
ifmarkets are perfectand complete, then ‡exible wages can undoallthe e¤ ects of…ring
costs and, therefore, …ringcosts are neutralon employment. W orkers pay ex-ante a fee
which is equaltothe severance paymentthey getin casethey are…red. Ifthey keep the
job, theygettheirfeebackwithhigherwages. Insuchaworld, foranylevelof…ringcosts,
itis always possible towrite an optimalcontractthatundoes allthe e¤ ects ofseverance
payments.
A thirdviewof…ringcostshighlightsthepossibilityof…ringcosts arisingendogenously.
T his approach is motivatedbythefactthatsometimes …rms andworkers negotiateseverance payments which donotcoincide with the ones legally set;oreven, some …rms o¤ er
severance payments in the absence ofemploymentprotection legislation. Severalauthors
haveinvestigatedthis ideain di¤ erentcontexts. Forinstance, B oothandChatterji (19 89 )
constructamodelof…rm-speci…ctrainingwhere thereturns totrainingareuncertain as
wellas theoutsideoptions forworkers. In such acontext, thecosts oftrainingareshared
between the…rm andtheworkerbecausethereexists thepossibilitythatworkers quit. In
caseofbeingdismissed, workers arecompensatedbythis costwitharedundancypayment.
A lso, B ooth (19 9 7 ) argues thatin atwo-period model, whereitis in theinterestof…rms
1
T he…rm has tocompensate theworkerwhen facingaredundancy whilenocompensation is required
in adisciplinary dismissal.
1
tohave long-term employmentrelationships, ifworkers are risk-averse they preferacontractwith redundancypayment, and risk-neutral…rms …nditoptimaltoo¤ erit. Finally,
Saint-P aul(19 9 6
) exploreshow…ringcostsariseendogenouslyinadynamice¢ciencywage
model. Firms maychose tovoluntarilyo¤ er…ringcosts in theirlabourcontracts because
these help …rms to credibly committo more stable employmentpolicies in an uncertain
environment.
T hese three views exposed above have radically di¤ erent ideas of …ring costs. O ne
couldsummarisecrudelythat…ringcosts are“bad” accordingtothe…rstview, “neutral”
accordingto the second vieworeven “good” accordingtothe third viewsince they can
be an optimalinstrumentfor…rms. Consequently, these three frameworks summarise all
possiblee¤ ects of…ringcosts overemployment. T hemodelpresentedhereis an e¢ciency
wage modelwhere dismissalcon‡icts arecostly. A s itwillbe seen, modelling…ringcosts
in this wayallows tointegratethethreedi¤ erentviews mentionedabove.
M ostofthe existing work on …ring costs focuses on labourdemand models and the
only type ofdismissals considered are redundancies2 . T hese models are very usefulfor
understandingthee¤ ects of…ringcosts on thedynamicfunctioningofthelabourmarket.
H owever, the e¤ ects on aggregate employmentareambiguous and remain in partialequilibrium. T he implicitassumption oflabourdemand models is thatwages are exogenous
and do notchange in the presence of…ringcosts3. In ourmodel, wages are endogenous
and …ringexogenous. In this way, the modelhighlights anotherdimension of…ringcosts
which is notcaptured by labourdemand models. T o focus on the e¤ ects of…ringcosts
on thewage-settingis particularlyimportantforthoseunemploymentmodels in which in
thelong-runtheunemploymentrateis determinedentirelybylong-runsupplyfactors (see
L ayard etal. (19 9 1)).
T here is a commonly held idea that…ring costs are high because dismissalcon‡icts
involve large administrative and legalcosts and thatthese lead to higherlabourcosts.
A lthough this pointis often made, itis usually modeled in asimplisticway: …ringcosts
paid by …rms are assumed to be higher than the indemnity that…rms have to pay to
workers4. B utthis is notactuallythecaseinmostEuropeancountries. Instead, thesource
ofhigher…ringcostshastodomorewiththefactthatthelegislationgenerallysetsahigher
severance pay forcases taken to courtand declared “unfair” than forthose considered
“fair” by court. T hese terms arede…ned from theworker’s perspective. A n “unfair” case
is when the courtconsiders thatthe …rm is wrongand thereforethe workermustreceive
the (“unfair”) …ringcostbecauseitis an unjustdismissal. T he“fair” severancepayment
is the defaultindemnity fora dismissal. W hen a case is taken to courtand is declared
“fair”, thecourtconsiders thatthe…rm is rightandtheworkersimplyreceives thedefault
2
See, forexample, B entolila and B ertola (19 9 0), B entolila and Saint-P aul(19 9 4), B ertola (19 9 0 and
19 9 2) and N ickell(19 7 8 ).
3
A n exemption ofthis is B ertola(19 9 0).
4
SeeB urda(19 9 2) foramodelofthis sort.
2
indemnity. Forinstance, in Spain5 , the costofa redundancy declared “fair” is 20 days’
wages pereach yearworked with a maximum of12 months’ wages. B utifthe case is
declared “unfair”, the costis more than double, 45 days’ wages peryearworked with a
maximum of42 months. T his “unfair” ratealsoapplies todisciplinarycases6.
In this paper, we explicitly modeldismissalcon‡icts and derive theircost. Con‡icts
between employers and employees can arise forvery di¤ erentreasons. In general, whenever…rms face a redundancy, they wantto use disciplinary dismissals in orderto avoid
paying…ringcosts. W emodel…ringcosts in acontextwhereworkere¤ ortis notperfectly
observable. T his actually provides a rationale forthe existence ofworkers’ rightto sue
theiremployers in case ofdisagreement. In such a context, workers can be in a weaker
position because …rms can getawaywith theuseofdisciplinarydismissals wheneverthey
need to adjusttheirworkforce (i.e. in case ofredundancy). T herefore, there is a reason
forjob protectionlegislationtoincludetherightforemployees totakecases tocourt. T he
drawbackisthatworkerswillthentendtodenyanydisciplinarycasetogetacompensation
(speciallyiftheindemnityis higherwhen thecaseis declared “unfair”), again, becauseof
the di¢culty in observingworkere¤ ort. A s itwillbe discussed, a double moralhazard
problem could arise. O urclaim in this paperis thatthe resolution ofthis problem by a
thirdpartywillbeimperfectgiven theinformation problem.
T heexistenceofimperfectresolutions ofdismissalcases willinturnimplythatdisciplinarydismissals willnotbecostless and…ringcosts willhaveanegativee¤ ectonaggregate
employment. A s itwillbe discussed attheendofthepaper, thesolution does notnecessarily imply the elimination of…ringcosts. R ather, whatwillappeartobe importantis
thegap between theseverancepaymentforcases considered“unfair” andthose“fair”.
W e concentrate on dismissalcon‡icts ofsmall/medium …rms forwhich “individual”
dismissalregulation applies. In case oflarge …rms, redundancies are generally underthe
“collective” dismissalregulation which implies thatthenumberofredundancies andtheir
totalcostarebargainedwith athird party(generally, unions)7 .
In ourmodel, …rms willbeara …ringcostthatis exactly the same as the indemnity
received by the worker. A s mentioned, this is the case formostEuropean countries. In
countries in which the administrative approvalprocesses are very complex8 , many ofthe
cases are settled by the workerand the …rm outofcourt, precisely toavoid these costs.
T herefore, again, …rms donotbearahighercostthan theindemnityreceived byworkers.
T he workerreceives asettlementwhich amountlies between thelegalseverancepayment
and the (expected) costhad the case gone to court. In this sense, the …ring costs due
5
See G rubb and W ells (19 9 3) and O ECD (19 9 9 ) fora comparison ofthese indemnities in di¤ erent
O ECD countries.
6
In this case, thedi¤ erencein costs ofbetween a“unfair” anda“fair” caseis extreme: 45 days’wages
peryearworked if“unfair” versus nocompensation if“fair”.
7
SeeB ooth (19 9 6and 19 9 7 ) foramodelof…ringcosts in unionisedsectors ofthe economy.
8
T ypically, this is casein southernEuropean countries (seeG rubb andW ells (19 9 3) andO ECD (19 9 9 )
forseveralindicators ofthe“strictness” ofemploymentprotection legislation).
3
todismissalcon‡icts thatare derived in the presentmodelcan be thoughtas the upper
boundofwhataworkercouldreceivefrom bargainingwith the…rm.
T herestofthepaperis organisedas follows. First, …ringcosts aredescribed, andtheir
implications for(e¢ciency) wages and employmentare derived. In the lastsection, we
discuss somepolicyimplications.
2
T he model
T he modelis a version ofthe shirking modelofShapiro and Stiglitz (19 84) with …ring
costs. A s in Shapiroand Stiglitz, a worker’s e¤ ortis notperfectly observable and there
is a detection technology that catches shirking workers (never erroneously) with some
probability q (where q < 1). W hen a workeris found shirking, he is …red and becomes
unemployed. W orkers alsohavean exogenous probability, b; ofbeingseparatedfrom their
job forredundancyreasons9 . Inthenextsectionwedescribehow…ringcosts aremodeled.
2.1
R edundancies and disciplinarydismissals incon‡ict
M ost industrialised countries have a job protection legislation framework that protects
workers againstredundancies. T heideais thataredundancyis anexogenous eventtothe
workerand imposes acosttohim and thus hemustbe compensated forit. A tthe same
time, employers areallowed to…reworkers fordisciplinaryreasons withouthavingtopay
anycompensation.
A frameworkwhereworker’s e¤ ortis imperfectlyobservableis bestsuitedforconsideringanothercommon feature ofjob protection legislation, namely therightforworkers to
sueemployers in caseofdisagreement.
W henever…rms needtoadjusttheirworkforce, theywanttousedisciplinarydismissals
toavoid paying…ringcosts. A nd the di¢culty in observingworkere¤ ortmeans there is
some chance that…rms can getaway with such strategy10 . T he rightforworkers to sue
employersincaseofdisagreementcancompensateforthisimperfection. B utthen, similarly
to …rms, workers willdeny any reasons fordisciplinary dismissalto geta compensation
based on unjust grounds. In such a context, both true disciplinary cases and hidden
redundancies arrivetocourtas disciplinarycases. Court’s decisions arebasedonwhatever
evidence (ifany) is presented bytheagents, which is notperfectlycorrelatedwith reality
given theinformation problem. So, in general, courts arenotabletoperfectlydistinguish
between true disciplinary cases and hidden redundancies. T hus, theresolution byathird
partywilltendtobeimperfectgiven theinformation problem.
9
T heterms adverse economicshocks and redundancies areused interchangeably here.
M alo (19 9 8 ) considers the case where …rms use disciplinary dismissals in cases ofredundancies in a
modelwhere…ringcosts arebargained in theshadowofthe lawbetween employerand employee.
10
4
Inthemodel, thisisrepresentedbythefactthatsome(true) disciplinarydismissalcases
could be mistakenly considered in favourofthe worker(i.e. “unfair”) and some hidden
redundancies could be mistakenly declared in favourofthe…rm (i.e. “fair”) by court. In
otherwords, in the…rstcase, workers arecompensatedwhen theyshouldnot. A ndin the
secondcase, …rms avoidpaying…ringcosts when theyshouldhavepaid them.
W ede…nem astheprobabilitythata(true)disciplinarydismissalismistakenlydeclared
“unfair”, where m > 0 given the information problem. T hat is, with probability m ,
disciplinarycases costthe“unfair” ratebecausetheyaredeclaredin favouroftheworker.
O nlywith probability(1 ¡m ) thereis nocourtmistakeanddisciplinarycases arecostless
tothe…rm.
W ede…nez as theprobabilitythata(hidden) redundancy is declared“unfair”, where
z < 1 given the information problem. T hatis, with probability (1 ¡z ), there is a court
mistake and redundancies arecostless tothe…rm. O nly with probability z , theworkeris
compensatedforaredundancycase.
Firms can betterprove thata(true) disciplinary dismissalis indeed disciplinary than
toprovethataredundancyis adisciplinarycase. T hatis, theprobabilitythatadismissal
takentocourtis costless tothe…rm ishigherwhenitisa(true) disciplinarydismissalthan
when itis a(hidden) redundancy, orthatz ¸m :In otherwords, theprobabilitythatthe
caseis declaredinfavouroftheworkeris lowerwhenthecaseis a(true) disciplinarycase.
W e assume thatthe legislation …xes a severance paymentofcforredundancies and
a severance paymentofC ifthe case is taken to courtand is declared “unfair”, where
c· C . T hen, given the double moralhazard problem, the …rm’s expected …ringcostof
a(true) disciplinary dismissalis m C and ofa(hidden) redundancy is z C . T able1 below
summarises …ringcosts described11 .
T able1: Firingcosts forredundancies anddisciplinarycases
R eality
D eclaration
of…rm
ofworker
R edundancy R edundancy accepts
R edundancy D isciplinary
denies
D isciplinary D isciplinary
denies
ExpectedCost
for…rm
c
z C
mC
T o sum up, given the contextdescribed above, there is always an incentive for…rms
todeclare redundancies as disciplinary cases and forthe workertodeny any disciplinary
case. A furtherdiscussion ofthis doublemoralhazard problem is donein thelastsection
ofthepaper.
11
L et’s assume thatin case ofredundancies presented as disciplinary cases, the …rm can nevershow
evidence ofthe case and the costis zC . In the case ofrealdisciplinary cases, ifthe …rm is able toproof
thecasewith probability k;then thecostis mC where mC ´0 (1 ¡k)+ kzC . T hus z¸m:
5
2.2
N on-shirkingcondition
In this section, we analyse the wageworkers mustbepaid in orderthatthey expend the
optimale¤ orton thejob. W orkers areriskneutral12 . T heirinstantaneous utilityfunction
is: U (w ;e) = w ¡e; where w is the wage and e is the e¤ ort. W orkers e¤ ortchoices are
discrete. Iftheyshirk, theyexpendzeroe¤ ortandproduction is zero. T hee¤ ortrequired
toperform in thejob is e > 0 .
W orkers choosethelevelofe¤ ortthatmaximises theirutilityactualised atrater. B y
i
V E; wedenotethepresentdiscountedutilityofanemployedworkerwhenshirking(i= S) or
non shirking(i= N ). Firms wanttoo¤ eracontractsuchthatworkers expendtheoptimal
e¤ ort. In whatfollows, the condition underwhich a workerwillchoose notto shirk is
studied(thenon-shirkingcondition, N SC ).
W hen aworkerdoes notshirk, hegets autilityequalto:
rV EN = w ¡e + b(V U + z C ¡V EN )
(1)
rV ES = w + b(V U + z C ¡V ES) + q(V U + m C ¡V ES)
(2)
whileiftheworkerdecides toshirkhis utilityis:
A s in Shapiroand Stiglitz (19 8 4), shirkingsaves the currentdisutility ofe¤ ortbutit
implies ahigherriskofbecomingunemployed. T his riskis proportionaltotheprobability
ofbeingcaughtshirking(q). Firingcosts alsoin‡uencethee¤ ortdecision herebecauseof
theimperfectcourtdecisions. W ith probabilitym ; shirkingworkers may becompensated
with aseverancepayment. T his reduces thecostofshirking.
T he workerwillchoose toprovide an e¤ orte; ifand only ifV EN ¸ V ES. W e can write
this condition usingequations (1) and(2) and gettheN SC in form ofutilities:
e
V ES ¡V U ¸ + m C ´K
(3)
q
T his condition states thatin orderto provide incentives, the punishmentoflosinga
job mustbeatleastequaltotheopportunitycostofshirking, denotedbyK . Substituting
this condition in equation (1), wegettheincentivecompatiblewage:
w ¸e ¡bz C + rV u+ K (r + b) ´ wb
(4)
In this wage equation, we can distinguish between the reservation wage (…rstthree
terms) andtherentlinkedtotheincentiveproblem (lastterm). ForC = 0 , this condition
is the same as in the originalShapiroand Stiglitz (19 84). In ordertoprovideincentives,
wages need toexceed the reservation wage by arent, K :T his rentis proportionaltothe
opportunity costofnotshirkingweighted by the term (r + b). T he higherthe discount
rate, the more a workervalues the savingofe¤ orttoday. T he higherthe probability of
12
Foramodelof…ringcosts where workers are riskaverse, see B ooth (19 9 7 ).
6
being…redforotherreasons than (truly) shirkingcases (i.e. shocks), themorecostlyitis
toexpend e¤ orttoday.
ForC > 0 , wecan distinguish twotypes ofe¤ ects of…ringcosts: thosedirectlyrelated
withtheincentiveproblem andthosethatarenot. Firingcostsa¤ ecttheincentiveproblem
becausetotheextentthat(truly) disciplinarydismissalsaredeclared“unfair” (i.e., m > 0 ),
legalseverance payments reducethepunishmentassociatedwith being…redwhen caught
shirking. T his implies that…rms havetopayhigherrents in ordertopreventshirking, as
can be seen in the above non-shirkingcondition (see equation (3)). T his e¤ ectof…ring
costs has thesame‡avouras thatintheinsider-outsidertheory, where…ringcosts increase
marketpowerofincumbentworkers.
A tthesametime, independentlyoftheincentiveproblem, theintroductionofmandated
severancepaymentsallowstheemployertoreducethewageexactlybythesameproportion
thatthepresentdiscountedutilityofanemployeeisincreased, withouta¤ ectingincentives.
T his can beseeninthe…ringcostelementofthereservation wage(seeequation(4)). T he
idea is thatlowerwages today, togetherwith compensation when being…red forshocks,
leavethepresentdiscountedutilityofbeingemployedunchanged. T hise¤ ectof…ringcosts
is thesameas thatproposedbyL azear(19 9 0)13.
A lthough this lastmechanism is notdirectlyrelatedwiththeincentiveproblem, ithas
veryinterestinglinks with e¢ciencywages in models in which …ringis notexogenous. A s
mentioned, in the standard e¢ciency wage modelwithoutseverance payments, workers
arepaid a“…ringpremium” in ordertopreventshirkingbecauseexpendinge¤ ortis more
costly the higherthe probability ofbeing…red duetoadverse economicshocks. W hen a
severance paymentis imposed, …rms face two opposite e¤ ects in the presence ofshocks:
they have to pay an implicit …ring cost to avoid shirking (the “…ring premium”), but
they can lowerwages because workers are beingcompensated when …red afterashock14.
A n importantfurtherinsightis madeby Saint-P aul(19 9 6): in adynamice¢ciency wage
model, itis in the interestof…rms to voluntarily include a severance pay in the labour
contractthatthey o¤ er. T his is onepossiblewayforthe…rm tocredibly committohave
amorestableemploymentpolicywhenfacingshocks, whichthenallows the…rm toreduce
directlythe“…ringpremium” tobepaid. T heoptimalseverancepaymentis suchthatthe
“…ringpremium” ofthee¢ciencywageis completelycompensated. In thepresentmodel,
the imperfectcourtresolutions imply that…rms donotwanttoo¤ erseverancepayments
toworkers15 .
13
So, form = 0 , thetwomodelshavethesamepredictions(seesection(2.4)wherethemarketequilibrium
is solved).
14
See Katsimi (19 9 8 ) fora more detailed derivation ofthis mechanism in a fully stochastic e¢ciency
wage model.
15
Ifm = 0 , …rms wouldo¤ er…ringcosts in thepresentmodel. Inthecaseofm = 0 , forC = e=qthetwo
models wouldcoincide. Still, in thepresentmodel, severancepayments aresetlegally whilein Saint-P aul
theyareendogenous. SeeB ooth (19 9 7 ) foradiscussionwherethelevelofmandated…ringcosts maydi¤ er
from those bargained.
7
B ack to the N on-Shirking Condition, ifa contractsatis…es the N SC , thatis, ifthe
workeris paid atleastwb or, ifbeingunemployedis asu¢cientlylargepunishment(V ES >
V U ), theworkerwillchoosetoexpendthee¤ orte. W ede…neV E as theexpectedutilityin
equilibrium. T he …rm chooses the minimum wage atwhich the workerwillnotshirk, so
thatin equilibrium theN SC is bindingandV E = V EN = V ES:
2.3 H iringdecisions
A ll…rms in the modelare identicaland in…nitely lived. T hey choseemploymentsoas to
maximise the expected presentvalueofpro…ts discounted atrater. W edenoteby ¦ the
presentdiscountedvalueofmarginalpro…ts. W ehave:
r¦ = f0(L) ¡w ¡b(z C + ¦ )
wheref(L) is theproduction function with f0(L) > 0 andf00(L) < 0 :
Inthepresenceof…ringcosts, themarginalcostofhiringaworkeris givenbythewage
plus thefutureexpectedcostofbeing…red. T hereis nocostofpostingvacancies, so…rms
hireworkers tothepointwherethemarginalpro…tis zero, i.e. ¦ = 0 . L abourdemandin
steadystateis given by:
f0(L) = w + bz C
(5)
T his equation shows that, forgiven wages, …ringcosts reducelabourdemand proportionallytotheirexpected presentvalue.
2.4 M arketequilibrium
Equilibrium occurs wheneach…rm, takingas givenallother…rms’wagesandemployment,
…nds itoptimalto o¤ erthe going wage ratherthan a di¤ erentwage. T he key market
variablethatdetermines …rm individualbehaviouris thepresentvalueoftheutilityofan
unemployed worker, V U . L eta be the rate ofexitfrom unemployment. T o simplify, we
supposethatunemploymentbene…ts arezero.
W ehave:
rV U = a(V E ¡V U )
G iven thattheN SC is satis…ed, wehavethatin equilibrium:
rV U = aK
(6)
N ow, substituting equation (6) in equation (4), we getthe e¢ciency wage curve in
equilibrium:
8
wb = e ¡bz C + K (r + b+ a)
(7 )
a(N ¡L) = bL
(8 )
Inequilibrium, theincentivecompatiblewageishigherthehighertheexitratefrom unemployment. T his resultis alsofoundinShapiroandStiglitz (19 84). T herentlinkedwith
theincentiveproblem isweightedbya becausethehighera; thelessbecomingunemployed
is apenalty.
W ederiveemployment, L, from thesteadystate‡owscondition: insteadystatein‡ows
tounemploymentare given by bL. O ut‡ows aregiven bya(N ¡L), whereN is thetotal
ofworkers in theeconomy. T hus,
T herefore,
L=
aN
a+ b
Combiningequations (5) and (7 ), we getthatthe equilibrium out‡owrate ofunemployment, a ¤, is given by:
f0(L) = e ¡bz C + K (r + b+ a¤) + bz C
(9 )
In equation (9 ), itcan be seen thatthe second type ofe¤ ectofseverance payments
mentionedbeforecan befullyundone: thesecondandthefourth elementofthis equation
cancelout. T heideais thatifmarkets arecompleteandperfect, and…ringcosts arefully
transferredtoworkers, then theyareneutralon employmentbecausethewageis reduced
bythesameproportion as theincreased shadowcostoflabour(seeL azear(19 9 0)).
H owever, in this model, even if…ringcosts arefully received by workers, they arenot
neutralbecausetheya¤ ecttherent, K . T hee¤ ects ofseverancepaymentsonthee¢ciency
wagesettinghavenocounteractinge¤ ects throughthenon-wagecomponentoftheshadow
costoflabour. T herefore, the wage schedule is shifted to the leftand ithas a negative
impacte¤ ecton employment. Itis interestingtonotethateven ifthewageis setby the
…rm, itis notpossible tofully endogenize the severance payments in the workers’ wage.
A s mentioned, this resultis due tothe presence ofa double moralhazard problem that
canonlyberesolvedimperfectlybyathirdparty. T his implies that…ringcosts haveareal
e¤ ectbecausetheyreducethecostofshirking.
T he aggregate N SC can alsobe written in terms ofthe unemploymentrate, u. R eplacingequation (8) intoequation (7 ), weget:
9
wb =
=
"
bN
e ¡bz C + K r +
N ¡L
#
e ¡bz C + K [r + b=u]
whereu= (N ¡L)=N .
T his expression can be represented in the (w ; L) space. Figure 1 shows the labour
market e¤ ects of neutraland non-neutral …ring costs. T he case where …ring costs are
neutralcorresponds tothecasewherethereis notsuch adoublemoralhazardproblem16.
A furtherdiscussiononthis problem andsomepolicyimplications is developedinthenext
section.
W
NSC
Ld
Non-neutral firing costs
No firing costs
Neutral firing costs
L(C=0)
N=L
L
Figure1: M arketequilibrium with neutralandnon-neutral…ringcosts
16
In such a case, disciplinary dismissals have zerocostand therefore they donota¤ ectthe renttobe
paid toworkers. Consequently, …ringcosts are neutralon employment.
10
2.5
P olicyimplications
In the modelpresented above, the existence ofjob protection legislation in a contextin
which workere¤ ortis notperfectly observable implied that…ring costs had a negative
e¤ ecton employment. A s explained, theproblem in such asystem is thatthereis always
an incentive forcon‡ictbetween employerand employee, i.e. forthe employerto claim
anydismissaltobedisciplinaryandfortheworkertodenyanydisciplinarycase. T his, in
turn, implies imperfectresolutions by third parties. T hegoalofthis section is todiscuss
somepossiblepolicyimplications derived from themodelabove.
T hemechanism thatgenerates suchdoublemoralhazardproblem is that, for…rms, the
expectedcostofaredundancyishigherthantheexpectedcostofdeclaringitadisciplinary
dismissal. A ndinturn, forworkers, theexpectedbene…tofdenyingadisciplinarydismissal
becomes positive. Followingthe modelpresented above(seeT able1), when …rms declare
redundancies as disciplinarycases andworkers denyalldisciplinarycases, thecourtis not
abletoperfectlydetectallthetruedisciplinarycases. T hecourtis abletocatch ahidden
redundancyonlywith probabilityz . A nditis abletodiscovertruedisciplinarycases only
with probability (1 ¡m ). T herefore, …rms …nd itworth itto declare disciplinary cases
when facingredundancies if:
c¸z C
(10)
m C ¸0
(11)
If…rms misusedisciplinary cases, then workers havean incentivetodenyanyofthem
because:
Ifthesetwoconditions aremet, then thedoublemoralhazardis an equilibrium. T hat
is, alldismissals aretakentocourtas disciplinarycases. A s shownintheprevious section,
insuchacase, …ringcosts arenotneutralonemployment. A s canbeseenfrom conditions
(10) and (11), policies thatconcentrate on undoingthe double moralhazard problem do
notnecessarilyimplythecompleteremovalofseverancepayments.
A s mentioned, in general, mostemploymentprotection legislation systems sethigher
severance payments for cases being declared “unfair” than for those considered “fair”.
T he idea behind this goes in the rightdirection in the sense thatittries to punish for
unjustdismissals. ForlargeenoughC , theincentiveof…rms tocheatcouldbeundone(see
equation(10)) andtherefore, therewouldbenodoublemoralhazard17 . Insuchacase, high
severance payments for“unfair” dismissals have a punishmentrole for…rms who would
use disciplinary dismissals when facingaredundancy. H owever, such apolicymaynotbe
su¢cient. IfC fails tobe high enough, itmotivates cheatingfrom both agents which in
turn generates imperfectcourtdecisions. A ndtheresultingaveragecostof…ringis higher
17
Ifthe …rm does notcheat, then the workerdoes notcheateithersince such strategy would be selfrevealing.
11
becausesomedismissals arepaidatthe“unfair” rate. M oreover, this does notseem tobe
the moste¢cientpolicy since itdoes nothave any punishmentrole forthe workerwhen
hedenies truedisciplinarycases.
A moree¢cientpolicywould beonethatpunishes any agentfoundlying. T hatis, on
theonehand, tosetaseverancepaymentthat…rmshavetopay, C F , whenthecourtcatches
ahiddenredundancy. O ntheotherhand, tosetapenaltyforworkers, C w , whenevercaught
denyingatrue disciplinary dismissal. T able2 summarises theexpected costs of…ringfor
the…rm and workerundersuch policyproposal.
T able2: Firingcosts: apolicyproposal
R eality
D eclaration
ExpectedCost
of…rm
ofworker for…rm forworker
R edundancy R edundancy accepts
c
R edundancy D isciplinary
denies
z CF
D isciplinary D isciplinary
denies
m Cw
U ndersuch apolicy, takingm and z as given, truth-tellingofboth agents is an equilibrium ifthefollowingtwoconditions aresatis…ed:
c¡z C F ·0
and
¡m C w ·0
N ote thatforany given m and z ; ahigh enough gap between severancepayments for
cases declared “unfair” and cases declared “fair”, thatis fora large C F ¡c
, and forany
positivepenaltytotheworker, thatis C w , theaboveconditions wouldhold.
T hispolicyhighlightsthatforanemploymentprotectionsystem towork, speciallywhen
workere¤ ortis notobservable, twothings areimportant. First, thedi¤ erencebetweenthe
levelof…ringcostssetforcasesdeclared“unfair” and“fair” hastobehighenough. Second,
di¤ erentindemnitiesshouldbesetfor“unfair” cases dependingonwhetheritis considered
thatitis theworker’s orthe…rm’s initiative.
3 Conclusion
Firingcostsareoftenblamedfordepressingemploymentlevels. B utthereareverydi¤ erent
viewsof…ringcostsamongeconomists: somemodelsindeedpredictthat…ringcostsreduce
employment while in other contexts …ring costs have no e¤ ect on employment. A lso,
12
in some otherenvironments, …ring costs are actually instruments chosen voluntarily by
…rms. In this paper, we have proposed amodelthatstresses thatitis notjustthe level
of severance payments what matters, but a wider view of employment protection. In
particular, dismissalcon‡icts and theircosthave been considered. A s discussed before,
themodelpresentedintegrates thedi¤ erentexistingviews of…ringcosts.
M oreprecisely, wehaveanalysedtheproblem behindthecon‡ictbetweenemployerand
employeein cases ofdisciplinary dismissals and redundancies, in acontextwheree¤ ortis
imperfectlyobservable. T hereis adoublemoralhazardproblem thatcan onlyberesolved
imperfectlybyathirdparty. T heconclusionisthat…ringcostswouldhaveanegativee¤ ect
on employmentbecause they modify the rentto be paid to workers in orderto prevent
thoseworkers from shirking.
T hemainpolicyconclusionsaretwo. First, tosetagap wideenoughbetweenseverance
payments forcases declared “unfair” and cases declared“fair”. Second, any agentcaught
lyingshould be punished. In ourmodel, di¤ erentseverance payments should be setfor
hidden redundancies declared “unfair” and fortruly disciplinary cases declared “unfair”.
W ithsuchapolicy, thedi¤ erent…ringcosts for“unfair” dismissals haveapunishmentrole
forboth employerand employee and, therefore, its implementation would eliminate the
doublemoralhazardproblem.
In this paper we have explored one possible reason behind imperfect court resolutions and theirimplication foremployment. T hatis, thefactthate¤ ort, which motivates
disciplinary dismissals, is notperfectly observable. T here are otherreasons why court’s
decisions could be imperfect. In ourcontext, thereis theproblem ofde…ningadismissal
caseprecisely. Forinstance, inthecaseofredundancies, itis di¢culttosetunquestionably
“howbad” the economicsituation ofa…rm mustbein ordertohavean objectivereason
to…reaworker. Similarly, fordisciplinarydismissals, itis di¢culttosetthedegreeofthe
worker’s faultthatjusti…es dismissal. T his leaves room forinterpretation implyingagain
thatresolutions willtendtobeimperfect. O therpossiblereasons arerelatedwithwhatthe
legislation considers as an “unfair” dismissalitself, somethingthatdi¤ ers across countries
(see O ECD (19 9 9 )). O r, howcomplex the dismissalprocedures are because this implies
potentially more strictproofs in ordertowin acase. In G aldón and G üell(19 9 9 ), we incorporatetheseelements inatheoreticalmodelandundertakeanempiricalanalysis totest
themagnitudeofdismissalcon‡icts andlabourmarketoutcomes fordi¤ erentcountries.
13
R eferences
[1]B entolila, Samueland G uiseppe B ertola (19 9 0): “Firing costs and labourdemand:
howbad is Eurosclerosis?”, R eviewofEconomicStudies, 57 , pp. 381-402.
[2]B entolila, SamuelandG illesSaint-P aul(19 9 4): “A modeloflabordemandwithlinear
adjustmentcosts”, L aborEconomics, 1, pp. 303-326.
[3]B ertola, G uiseppe(19 9 0): “Jobsecurity, employmentandwages”, EuropeanEconomic
R eview, 34, pp. 8 51-886.
[4]B ertola, G uiseppe(19 9 2): “L aborturnovercosts andaveragelabordemand”, Journal
ofL aborEconomics, 10 (4), pp. 389 -411.
[5]B ooth, A lison L . and M onojitChatterji (19 89 ): “R edundancy payments and …rmspeci…ctraining”, Economica, 56(224), pp. 505-21.
[6
]B ooth, A lison L . and A ndrewM cCulloch (19 9 6): “Firingcosts, U nions and employment”, CEP R discussion paper 1347 .
[7 ]B ooth, A lison L . (19 9 7 ): “A n analysis of…ringcosts and theirimplications forunemployment”, in D ennis J. SnowerandG uillermodelaD ehesa(ed.), U nemployment
P olicy: governmentoptions forthelabourmarket, Cambridge: CambridgeU niversity
P ress.
[8]B urda, M ichaelC.(19 9 2): “A noteon…ringcostsandseverancebene…tsinequilibrium
unemployment”, Scandinavian JournalofEconomics, 9 4 (3), pp. 47 9 -489 .
[9 ]G aldón, Joséand M aia G üell(19 9 9 ): “Firingcosts, dismissalcon‡icts and employment”, P rinceton U niversitymanuscript.
[10]G rubb, D avid and W illiam W ells (19 9 3): “Employmentregulation and patterns of
workin EC countries”, O ECD EconomicStudies, N o. 21, 7 -58.
[11]Katsimi, M argarita(19 9 8 ): “Firingcosts ande¢ciencywages”, mimeo.
[12]L ayard, R ichard, Stephen J. N ickelland R ichard Jackman (19 9 1), U nemployment:
M acroeconomic P erformance and the L abour M arket, O xford: O xford U niversity
P ress.
[13]L azear, EdwardP.(19 9 0): “Jobsecurityprovisionsandemployment”, Q uarterlyJournalofEconomics, 105, pp. 6
9 9 -7 26
.
[14]L indbeck, A ssarand D ennis J. Snower(19 8 8): T he Insider-O utsiderT heory ofemploymentandU nemployment, Cambridge, M A : M IT P ress.
14
[15]M alo O caña, M iguelA . (19 9 8): L as indemnizaciones por despido, un problema de
negociación, M adrid: EditorialA CA R L .
[16
]N ickell, Stephen J. (19 7 8): “Fixed-costs, employmentand labourdemand overthe
cycle”, Economica, 45, pp. 329 -345.
[17 ]O ECD (19 9 5), T he O ECD Jobs Study, P aris.
[18]O ECD (19 9 9 ), EmploymentO utlook, P aris
[19 ]Saint-P aul, G illes (19 9 6
): D uallabourmarkets, a macroeconomic perspective, Cambridge, M A : M IT P ress.
[20]Shapiro, CarlandJosephE. Stiglitz (19 8 4): “Equilibrium unemploymentas aworker
disciplinedevice”, T he A merican EconomicR eview, 7 4 (3), pp. 433-444.
15
CENTRE FOR ECONOMIC PERFORMANCE
Recent Discussion Papers
462
W. H. Buiter
Optimal Currency Areas: Why Does the Exchange Rate
Regime Matter?
461
M. Güell
Fixed-Term Contracts and Unemployment: An Efficiency
Wage Analysis
460
P. Ramezzana
Per Capita Income, Demand for Variety, and International
Trade: Linder Reconsidered
459
H. Lehmann
J. Wadsworth
Tenures that Shook the World: Worker Turnover in Russia,
Poland and Britain
458
R. Griffith
S. Redding
J. Van Reenen
Mapping the Two Faces of R&D: Productivity Growth in a
Panel of OECD Industries
457
J. Swaffield
Gender, Motivation, Experience and Wages
456
C. Dougherty
Impact of Work Experience and Training in the Current and
Previous Occupations on Earnings: Micro Evidence from
the National Longitudinal Survey of Youth
455
S. Machin
Union Decline in Britain
454
D. Marsden
Teachers Before the ‘Threshold’
453
H. Gospel
G. Lockwood
P. Willman
The Right to Know: Disclosure of Information for
Collective Bargaining and Joint Consultation
452
D. Metcalf
K. Hansen
A. Charlwood
Unions and the Sword of Justice: Unions and Pay Systems,
Pay Inequality, Pay Discrimination and Low Pay
451
P. Martin
H. Rey
Financial Integration and Asset Returns
450
P. Martin
H. Rey
Financial Super-Markets: Size Matters for Asset Trade
449
A. Manning
Labour Supply, Search and Taxes
448
S. Nickell
L. Nunziata
Employment Patterns in OECD Countries
447
W. Buiter
The Fallacy of the Fiscal Theory of the Price Level
446
R. Portes
H. Rey
The Determinants of Cross-Border Equity Flows
445
G. Duranton
D. Puga
Nursery Cities: Urban Diversity, Process Innovation and
the Life-Cycle of Products
444
D. Quah
UK Phillips Curves and Monetary Policy
443
L. Feinstein
The Relative Economic Importance of Academic,
Psychological and Behavioural Attributes Developed in
Childhood
442
D. Quah
Cross-Country Growth Comparison: Theory to Empirics
441
D. Quah
Internet Cluster Emergence
440
E. Mellander
The Multi-Dimensional Nature of Labour Demand and
Skill-Biased Technical Change
439
F. Green
A. Felstead
D. Gallie
Computers are Even More Important than You Thought:
An Analysis of the Changing Skill-Intensity of Jobs
438
M. Güell-Rotllan
Workers’ Transitions from Temporary to Permanent
Employment: The Spanish Case
437
S. Redding
The Dynamics of International Specialisation
436
M. West
M. Patterson
J. Dawson
S. Nickell
The Effectiveness of Top Management Groups in
Manufacturing Organisations
435
F. Green
S. McIntosh
A. Vignoles
‘Overeducation’ and Skills - Clarifying the Concepts
434
H. Overman
D. Puga
Unemployment Clusters Across European Regions and
Countries
433
D. Puga
Diversity and Specialisation in Cities: Why, Where and
When Does it Matter?
432
S. McIntosh
A Cross-Country Comparison of the Determinants of
Vocational Training
To order a discussion paper, please contact the Publications Unit
Tel 020 7955 7673 Fax 020 7955 7671 Email [email protected]
Web site http://cep.lse.ac.uk