PSfragreplaements
0
0 L
w
F −1 (g) L
ritp = 0 w
ritp → ∞
Ω = 0 Π = 0
Figure13: Employmentwage pairs underindividual rationality and feasibility
of the prie funtion is zero and hanges sign, and where the prie and prot beome
in-nite. Thus, the set of bargaining pairs
(L, w)
with positive prot onsists of the union of twodisjoint open regions allowing unbounded wages for
L < L
rit and unbounded employmentlevels.
15
As aonsequene one nds that the set of individuallyrational and
inome/demand-onsistent employmentwage pairs takes the form of a union of two adjoining sets as depited
in Figure 13. Observe that the two ritial employment levels, whih are the same for eah
stateof the eonomy
(M, p e )
,are determinedby demand featuresand theprodutionfuntion.They are independent of money balanes. However, high prie expetations may make the
lower ompat urvilinear triangle empty, implying that all equilibrium alloations must be
in the upper region of feasibility. Sine unbounded wages with unbounded pries are feasible
inome/demand-onsistentequilibriumalloationsforemploymentlevelsneartheupperritial
level, the assoiated set of payos must beunbounded and be equalto allof
R 2 +
.By adding the equilibrium points and the
λ
-eieny frontier to the above diagrams, oneobtains inFigure 14a omparison of all senarios in alloationspae and inpayo spae. For
the isoelastiexample,allequilibriaareinthe ompattriangular regionofthe employment
wage spae. This shows also that the two one-sided strategi monopolisti situations indue
ineientemploymentlevelsbelowthe eieny frontier (leftpanelof Figure14). In ontrast,
theomparisoninpayospaeonrmstheloationofthetwoone-sidedmonopolistiequilibria
above the
λ
-bargainingfrontier, see Figure14(b). In otherwords, both monopolistiequilibria induebetter payoswhihannotbereahedorsupportedby theooperativedeisionsundereient bargaining. Notie, however, that the union's payo for
λ = 1
is less than at thenonooperativeequilibriumwhiletheproduer'sprotishigheratthenonooperativesituation
thanunderbargainingwith
λ = 0
. However, theserelativepositionsofthepayosdependontheprieexpetations. AsFigure15shows, thepayosinbothnonooperativeequilibriaarehigher
thanthemaximalpayosunderbargainingwhenexpetedpriesarehighenough. Theloation
in payo spae issurprising and ounterintuitive at rst. The arguments disussed at the end
of the two monopolisti ases show that, for eah given prie level
p
in the nonooperative 15Stritly speaking, the set also ontains the boundary point
(L
rit, w
rit)
sine there exists an unboundedintervalofpositiveprieswhihindue positiveprots.
PSfragreplaements
0 0
om
λ = 0 λ = 1
L w
F −1 (g) L
ritp = 0 w
ritp → ∞
Ω = 0 Π = 0
union
mon
(a)employmentwagepairs
PSfragreplaements
0 0
om
λ = 0 λ = 1
rit
rit
union
mon
Ω
Π
union
mon
(b)payos
Figure14: Wages, employment, and payos under low prie expetations
situation,the monopolistan exert market power to obtain the full rent fromthe ompetitive
agent, a possibility whih neither the union nor the produer an obtain under bargaining.
Thus, the prie feedbak seems to wash out this eet under ooperation.
PSfragreplaements
0 0
union
mon
om
λ = 0 λ = 1
L w
F −1 (g) L
ritp = 0
w
ritp → ∞
Ω = 0 Π = 0
(a)employmentwagepairs
PSfragreplaements
0 0
union
mon
om
λ = 0 λ = 1
rit
rit
Ω
Π
(b)payos
Figure15: Wages, employment, and payos under highprie expetations
The diagrams are drawn for the parameters of Table 1 and given values of the government
parametersandforgivenvaluesofthestatevariablesmoneybalanesandexpetations. Beause
of ontinuity, these features are loally robust properties and they will be observed for this
isoelasti lass of models in dierent magnitudes and possibly also in dierent relative orders
under dierent parameters and values of the state variables. However, as some numerial
experiments have shown, the basi features are preserved for a wide range of values of the
parameters and of state variables. The overall homogeneity of the prie law and the wage law
does not prelude reversals oropposite eets.
While these result might seem to be ounterintuitive at rst sight, it is straightforward to
disern the two prinipal reasons why these eets our. First of all, the maximization of
nominalobjetives(protresp.exess wages)reatesspilloversbetween markets even forstati
general-equilibriumsystems,whihareprimarilyduetoinomeeets. Beauseoftheseinome
eets, it isunlikelythat the universal omparative-statisresults (asoftenderived in
partial-equilibrium models with strategi behavior) will persist in general-equilibrium models. It is
knownfromgeneralequilibriumtheorythatsuheetsareduetoprienormalization,implying
dierent real alloations, relative pries, and nominal values of inomes (prots and wages)
underdierenthoiesofanumeraireorofprieindexes. Theseresultsarewelldoumentedand
havebeenreognizedinmanydierentontextsinpartiularinwelfareeonomis,international
trade, or oligopoly theory whenever inome feedbaks are taken into aount appropriately
with a nononstant marginalutility of inome for onsumers.
16
In temporary equilibriumof a
monetary eonomy, these eets learly do not disappear.
Seond, the prie feedbak, whih was shown to be responsible for the ineieny of the
bar-gaining solution under ompetitive prie taking in temporary monetary equilibrium, operates
in eah of the three ases endogenously in a dierent way. There is no strutural feature of
the modelwhih relatesthe nominal payos, hosenfor the bargainingproblem neitherto the
nominal objetives by the monopolist/monopsonist with wage setting and prie taking nor to
the results indued by the maximizationunder ompetitive prie and wage taking. Thus, in
all three ases, the prie feedbak and the inome feedbaks have a deisive inuene on the
nominal values hosen for the payos in the monetary eonomy. For these reasons, the four
labormarketsenarioswhose equilibriumharateristis areompared inthepriewage spae
and in payo spae are in general not omparable with respet to real alloationsor nominal
payos, even under the weak onept of eieny. Sine, in addition, equilibrium pries and
alloations depend on the other state variables, anextensive welfare analysis may not lead to
onlusive results.
It is worth noting that some properties of the results are spei to the isoelasti model
ho-sen for the numerial analysis sine the bargaining parameter
λ
plays a spei dual role intemporary equilibrium. On the one hand, there is no impat of union power on aggregate
supply. Therefore, the interation of the isoelasti struture between prodution and labor
supply shows that the measure of union power
λ
exerts a diret inuene on the real wagemark-up and on the level of underemployment, making both of them onstant in temporary
equilibrium. Theseonstantsdepend ontheelastiitiesof thelabormarketpartiipantsandon
union poweronly. Thus, in a dynami eonomy as analyzed inthe next setion, both of them
are onstant over time, i.e.independent of
(M, p e )
, and they are independent of allsal and demand parameters in the eonomy. On the other hand, a powerful union whih an hoosethe parameter
λ
doesnotexert absoluteontroloveritsseeminglymost importantendogenousvariable the wage rate. Moreover, even for the isoelasti ase, it seems unlear whether the
wage outome under bargainingdominates the ompetitiveoutome, insome other sense than
theeieny riterionusedabove. Itremainsanopenquestiontowhatextenttheineienies
will hange or disappear if the bargaining agents hose real rather than nominal payos as
objetives.
16
seeforexampleDierker&Grodal(1986);Böhm(1994);Gaube(1997);Roberts&Sonnenshein(1976)
5 Dynamis of Monetary Equilibrium
So far the harateristis of equilibria under bargaining were disussed for an arbitrary given
period
t
with initial money balanesM t
held by the private setor, expeted pries for thenext period by onsumers
p e t,t+1
, and by the union powerλ t
. Thus, the triple(M t , p e t,t+1 , λ t )
desribesthe stateof theeonomy atany given time. Assoiatedwitheahstate arethe pries
and wages and the levels of output and employment
(p t , w t , y t , L t )
in temporary equilibrium whihare dened by applying the respetive mappings fromthe previoussetion.17
Thissetionanalyzesthedynamibehavioroftheeonomy inequilibriumassumingthatunion
power is onstant over time and given exogenously at some level
0 ≤ λ ≤ 1
. Sineλ/(1 − λ)
determines the relative share of wages over prots, no other eonomi variables related to
the objetives of the agents are onsidered. As was shown in the previous setion,
λ
has asigniantimpatonmost importanteonomi variablesineveryperiod,likeoutput,inomes,
pries, and onsumption, whih are relevant for welfare. Thus, it would be desirable to relate
thespeivaluehosenforunionpowertothemarketdatawhihareinduedandtoreevaluate
the equilibrium outome with respet to the true objetives of the agents. This leads to an
endogenous determinationof the measure of bargainingpower. Forthe dynamis, this implies
that an adaptive rule or a dynami mehanism has to be dened based on the data in eah
period. However, at this stage we examine the dynamis of the monetary eonomy without
providing any justiation what level of union power
λ
would be reasonable to be assumed,leavingsuhquestionstobeaddressedinfuture researh. Therefore, the dynamidevelopment
of the eonomy will be desribed ompletely by haraterizing the evolution of the two state
variables money balanes and expeted pries
(M t , p e t,t+1 )
, implying a two-dimensional state spaeX := R 2 ++
.5.1 Perfet Foresight
A sequene
{p e t,t+1 , p t } ∞ t=t 0
of pries and expetations will be said to have the perfet-foresight propertyifp e t,t+1 = p t+1
holdsforallt
. Itisoneofthemainquestionsofdynamimaroeonomianalysistondonditionsanddenetheoneptswhihensurethatperfet-foresightsequenes
are in fat generated by an assoiated dynamial system whih is globally dened. In other
words, a foreasting rule ora preditor has to be dened to ensure perfet foresight alongany
orbit.
18
In order to guarantee that, for any period
t
, the atual priep t
oinides with itsassoiated predition
p e t − 1,t
, the onditionp e t−1,t = P (M t , p e t,t+1 , λ)
must hold for any
t
. This denes impliitly the funtional relationship determining how the foreast inany periodfor the next one should behosen as afuntion ofthe previous foreast.Therefore, solving (16) for the expeted prie
p e t,t+1 = ψ ∗ (M t , p e t − 1,t , λ) ≡ P e M t , p e t − 1,t , λ
:= p e t − 1,t AS −1
D M t
p e t−1,t , λ
17
We will assume throughout this setion that the aggregatedemand funtion is independent of expeted
ination. ThegeneralaseouldbedealtwitheasilyusingtheresultofLemma3.1.
18
seeBöhm(2010)
denes the perfet preditor
ψ ∗ (M t , ·, λ)
sine forall(M t , p e , λ) P(M t , P e (M t , p e , λ), λ) = id (M t ,λ) (p e ).
Therefore, the two mappings
M t+1 =M(M t , p e t − 1,t , λ) := M t + p t (g − τ D ˜ (M t /p t , λ)) p e t,t+1 =ψ ∗ (M t , p e t−1,t , λ)
(50)
with
p t = P (M t , ψ ∗ (M t , p e t−1,t , λ), λ)
and˜ τ ≡ τ ˜
p e t,t+1 p t
, λ
= ˜ τ
ψ ∗ (M t , p e t − 1,t , λ)
P (M t , ψ ∗ (M t , p e t − 1,t , λ), λ) , λ
denethedynamibehaviorofmoneybalanesandexpetationsunderperfetforesightforany
level of bargaining power
λ
. In addition,τ ˜
denotes the average tax rate whih willbe derivedin(52). Sine for all
t
, one hasp e t − 1,t = p t
, one an rewrite (50) asM t+1 = M(M t , p t , λ) =M t + p t (g − τ D ˜ (M t /p t , λ)) p t+1 = ψ ∗ (M t , p t , λ) =p t AS −1
D
M t
p t
, λ
,
(51)
dening equivalent dynamis with perfet foresight in the spae of money balanes and pries
(M, p)
for any given levelλ
of bargaining power.It is one of the reurring themesof dynamialeonomies with prie expetations that in most
ases prie dynamis indued under perfet foresight are unstable, a phenomenon whih also
oursintheurrentmodel. Toseethis, let
M > ¯ 0
denoteanarbitraryonstantlevelofmoneybalanesand
λ
begiven. Then,(51)redues tothe one-dimensionaldynamialsysteminpriesG : R ++ → R ++
,p t+1 = ψ ∗ M , p ¯ t , λ
=: G (p t ) .
Rewriting (16) one nds that ithas the unique positive xed point
p = M ¯
D − 1 (AS(θ e ); λ) = M ¯
D − 1 (AS(1); λ) ,
where