Professional Documents
Culture Documents
tsARGAIININC
Edired by
Ken Binmore
and Partha tr)asgupta
Basil Biacki¡,e11
Cc.pyriglit t' Ken llinmore an¡.l Partha t)*sgupta l9g7
Reprinted 1990
N ash Bargaini*g'Fheoly:
An Intreiduction
K. Binntore and P Dasgwpta
i BACI{GROUNI)
i
l¡ spite oi'early hopes, it is oniy in recent years lhat the von l'ietl¡lann anrl
Vlr)rgersiilrn iitt¡or!' ol garnes has begun to be gerrr.linely fnriLl'ui in economic
x :rn:r1r,-sis. Lli r*trrlspect, it s*ettrs clear that lhe delay rvas due cr':iy in iiarl 1o
¡he inco:rri:lriilness of the theory'. since nri.¡ch of what is nor¡,, L'*ing used iras
be::l a.;ei]¡b1e in essence iic¡nr ihe eariy 1950s. A larEer stuirrbling biock lias
Lrecn tii.t ¡r:.r:ble¡r of cletermining under what circuinst:lnces tiie avaiiable
throrf is {ur is not) applicabie.
¡larne'llrer:rreti": anai,vsis, one is fai:i'd rvith tr'¿o
"tVhcn seeliing to err.lploy a
iunilanlcrt tlI questions.
:l'"1;,j;ti,;'::,j,::',|,X";";.;j;;*'rT' !n ihe ntrr]r"uctioi'l tll tlie companiln vi:lurne' r've intio''iu;e': the idea oi a
(iwttts"was irlsis{ent 'rf i]r.ir-rt i(l n iar i11e conip:1ni': ir. v':iline.)
"t';l;--;;1'l;f'¿'1¡¡
Llutthi:;idearronlycnctr,ilev*raiinrpor;ientcor;iri''*ul!i;'lising¡J;i:'.'"..ashl
tit**i f"tlt-ps ;rri ídc'a rri ':;-'i.lxiil wlt*rr p0sii'nle. ir,¡- e:{.4fip1{. l:riay¡¡,, L:f i.i ¡t:ltt'lt lviil r.l'*ulll!' "t'isii ttl
t{1 sall}*
c1-Lrri*¡l the shor.t tirnr-'
lle ti*vr¡te*l :...:i¡tdir:A;e tireir Strat*uy Ci-tr-.ic.e¡.a If ¡r...i1]i: Ol,i.ti:¡t 3i11 relUClitill lo t{-li]iit'rAtil
iltie¿i1*:ileri. li s.r, horv Lrig shi-rulil tii¡':
'*bii:'li it is pix-l';:'; itr'i n,:ri:,i¡::; f-it.;-l' .,lr¡uld be bribeii oi
inrpotlancelscn.l,¡¡,*.:ln*.hat1'*.,no,,.b*cc,iilc!.ilt]r,"¡11..'qti¡¡,Jd..,'.r:¡l.j
"[']ris provides *''""ttnt*'f tr¿ll':rvor'ii witirin
3,ra;;r'.
evaluaiegener',rlu""n''p"on'about.optrn-i*ft'*tt'"i'tt''ina'¡nlfi;'"riiLlli il .X.i. lit¡ i,-\¡?!-tl)i¿. t: sii*¡rc:i'i*ri e.rp!:i'.-'ii1.,,1.1'l::lritr:t.:.i;1.* liís ot!--e¡$¡sa irt:!1Itt{ti]{
o tlt"tt't of soluii:'r' i:',.n'1
1i¡r'rdamc*:*i P,.lili:-:: tit't
riiscrpirneCr n1¿iIner' The ir¡lit ri'i7i).
I l\'ilelc ¡]1¡ ¡r¡r¡rorii.rrrii:c:l lcr rle"¡llay;]f:ilii.\'('!irsn iü thc p[¿]e¡s ure sulficilntly
ceptforag¿llll''the(lreticall-vbaseri*cono¡"it'ntti*iiiterlslober';]lrl':'lrerlj 1 : j;i¡ ;i r'an l::¡:pi';; th:i: ül: ;:;i:::cise ,ritura ol '.1i* fi¡r¡¡r¿/ st:ll:':gies
1i)¡ L; t--'cirilrif c ;:1m{}s1
ii¡cicrant l-re5!¡ip1.i*n1ir,: iitl ir-)!nlal st¡ategie:; ale tht'tt tlpicallv t¡r:litte¿i
ll,,ll-1.,,";;;.'iii'.:ilü:'::':.'iiJ"l'".lT-;L"'::":$' ,l As in ih¿ lrii.l! r;i ¡:i¡¡-¡,:i;jc,.J *quilíbria {!*r Jll!'eIlü!l 198"1 l'c)¡ a lopical e.a(}lint}
,i,Jx;;:;:'il;';lT.'';l'':;; t'"t w*itt"u¡ l9?5' p' 70)' I
l.,ossibic'd!ililor1lir
ou"trnt"'i
theo4': &n ifrlrcld!"$:ti{}f, l{as;k hargaining thettr-v; an introductiotr
.\¡as& bcrgcerr ing
irrihr's Lri:'l \1'!rer threats will be ellectivril Sui:h pre-pla¡¡ aciivit-v is nÜi always
8r{bre strategies are L:hJ$en i'a rormar ganre piayeci r;ooperativ*iy. !t !s
pla-ved by two con- sui]!'osed tilat ¡:rla.yers m0y commufiicate cerstiessly and without rijs:¡lcti.}¡r
¡ieriiriiled. ljor example, bribeLy in tl.r* speeies $[^ganie or.i tilut thel'luay eiiter intc'r aily agreenients whatsocver that they ühoose.
irr{,,,.,r, srrblnitrLfig tendsfs t() a public ollicial is illegal. Sinlitrarly, ¿nti-trtlst
i¡:rv:r ¿re designed to preve$t tlte fornlatio¡t of certain coaliiions for nrutu¿l i¡l all, it is assumed ti'rat inechanisms eyolrt fr:r eritbrcin¿ sucS
.\,!ost irriportant
is the question crf \\'hether any as'eenrfnt$^ ¡\ side paylllent is to be thoughl *f as a su¡ri of mcneY whicir
ac\,:il1rage P,,rliaps of most vital irnportance
on the players. The i]lterest of tire parlour ;;iay be paid bi' one player to another in or<ier to facilitate ('r liincer) rht:
pÍc-iri;iv asreen.!ents are tsintling
sigiring r:f a pre"play cont'act. In principle, rhe notion of a non-cc¡*perari\¡e
g*n¡"' .Di¡:ior¡ecr", for example. lies i' the judicious breakirig r-rf agreemenis.
gaine covers a wide speclrum ol'possibilities but attention is usually
*l:ile. lii the c'¡mnlercial worltl, contracts are supposedly legally e¡tforceable' arjdressed
to rvliat Flarsanf i c¡rlls'tacit games' and r"ve prefer to call contests" It is to be
l)n ¿ ¡rror* fuüdamental level, it rnay be that th€ pia-vers cannot e ommunicate
witl¡ *ach other at ali (e.g. if one does not knorv lhe other's lvhereabotits) oi
r¡nderstood that, befi¡re the play oi such a gan¡e, the
¡rlayershave no ijpp,.]r-
¡unities. explicit r"rr impiicit, for an1' t,vpe of comnunication at all.
th;:r i]re+tii.:gs of tl-rree or rnore players are impossib!e {e.g. if t}rey coil,mui.ti-
cat* Liv telephonel.s Where self-bintling threats are possible, the or'1er üt
Thi,s cls5sifi,;utri-¡¡t rs tor; crucie to prnvide more than prelirninilr¡,
assistanr*
iir e';aluati'g solution co¡¡cepts. For this reason ce;:tai. coücepts fbr cc,;pera-
uili¡!i tlie plal'rrs can ¡rrake their statements becornes highlV relevant' T'his
niade physic- tiv* games come equipped rvith characterizing'axionl systsms'. chapf*r I.f
bri;tg.s us to ihe gerre rai question of corn¡nitment. conllnitments
ali¡ {iclr example. by investing'in a pia*l} are easijt nrodelled;but what of iiiis voiume, lor example, provides an alte¡native axiorn svst*¡n to thatusualiv
extent lio these binct those: who make thei¡'l luiit¡red in su¡rporl- o!' ¡he Nash bargaining soiution. But siich axiom svsr*r¡¡s
v.:i'c:1 con-i:¡it¡l:egts? To what
the players make ¡lre credible because ol ;".' :r'pic:rlly couchcd in ratjrer abstracr tenns and rllere s*,:m t.¡ be r:,-r u,,:i,s
uar¡ or,e l'L{¡lfioslj tliat cornmitments
to the colnñit|nfnt .¡'ouid irave ern r']rei¡ rt ¡.li whe re the appropriate interpretaiion is not controversial.
rire unpaci tltat a failure keep
'i'h,': Nash proLlraril (outlined
rep¡irerlofis? so. what is the mechanism? {For a. irofe rielarieci iliscussi"rt
lf in a itiv sentences by l"{asir ig5i) prcr.i,"lcr a
relsteii questions on trLlst' see Dasgupta 19'56 ) much r¡o¡e satjsfact*ry franiewr:rk withi* which r,: *nnuid*r ,*.,*,,
c,f r.hese and 'i¡r'
ir'iiticr$. Ili"c ii:rsle vierv rvas that the nrost funilamental tn¿le olgarne is g.hat
Tl-rese erorrsiderxiions are iisted in an attenrpt t'.¡ r"ni,ike it clear
ttr:!t wh3.t
or ffii:lY not l.i3ii' '¿'; herr: c¿11rd :r r'olttesÍ. We l-rave
happtrits 1It a gaine rvi.trtr in general depenti r]c \r'hat n1üy disciissrri tir* lrilre-the omtic anaiy..:is of sr-¡ch
trapper.e¿ befule úe fornrul ganie is plaveel {.or *n ""'hat is anÍi;ipaleduiigiLl
i;r:rer, 1il fhe i:ornpanioir voir.u.r'l*. Brieliy. lhe r¡otion cif a ,,vss& equilibriutzt
¡:l¡i:- pro!ides e i"irrn fbundatli.r:i r.lpon rvirich the tireon'c.rii rrst.. j'hi:;;s
hopr*,,I *,i,:er the ibr¡nai garne is plal'eri). But strlr¿"iclt cirllc!.'pti sirr'plY not tú sat.
sisi ol r:.rappii',gs frcrm a fcrmai game Úi (often rr:ppr'l-i;lüg' 'oiich .'ii 1:ire Itr:[ prohlc:irs ijo not e;i;sr nr:r xjial r*finemcnlj. oi'¡hr hia:h ei;uiiibiium itje:r
sln:ifure uf {r-,¡ r,,r outcomes of C. It ft¡llows thal rl:e ii:-*i|rc t¡i a rolutii-'n air iril¡ r:s¡entlai.6 {Jn tii¿ coutrary. there are rlumerou$ eiil.ficuit¡es to whirir
ire sh¡ii i'ini i.r n,:crssftrv to ieti-ri!). Ilut thesr: diificultici arc ¡.,1 n secoiriii,,
collillrt ntusr necfssariiy entail a whole nlix of inlgilicit assutttpiiolis: abr''tli
íi¿itufe cí;\rnpa¡eil lvith tire crnceptuai prcble'ms ¡esoJu-*J f0r c0ni3:iis
lhe n:ture of iile iooperative infra:;tri¡cture r','ithin ¡¡'ir:r:h tlie p'aru* is tho''rpht hV ll;,;
i,l,::l t-ri'a }rj¡sll *qi-iil.ibriuirr hut,.i,htcl-i remain la:"ge11.,Llriii.r-qoiv*d fo: r';lir,:,1
t{) ri.]iriate- ¡\nc. ro (:arry the same point turther. .h. er:istenre trf :, ba't¡en'
ivpes ofganre.
ol su|'ii soir¡tion aoflcepts necessarily implies tire eristence ol ¿ C/dss¡Íic¿¡riii¡¡
we nol'r' turri !.o qanrer clhei firari contcsts - i.e. gl;;e:i ,¿'liicir ai.e i,r i r,
inirastructures - each ¡iilferent concepi l:;ring aÍ:pr,r¡:iriale f Jr
c,.r,,-,¡,eriiive
'f¿,-1iite rent iiti¡astn¡cture lln,rl,r's¿d on thtl r¿ss:-;r111:;ticit iii.*t sJms tneasurr;,ti r:rr:-¡;iLrl,inle ractier¡
btlr. Jr,.i
"
is alrn*si iiracrt,u.li' ilre ¡:i¿ys¡s is pr',s:ible.j i\j¡¡rb p¡rr¡¡115sd tr: deai ",¿-iiir the
'{
}.,e stanijaril ciassilication. illustrated Lrelow" ol pr*-¡:ri:r.,
¡;rc,l;lerrr
pl'1liiií11/ e
ír.tivi1y irl th¡: i'oiii:'xing vu,¡i¡. it G is a l¡¡r;r;l {alíu.:. it¡lr::ine tlrs vlr¡r._r¡-¡:,
¡rrlssil,rle steps !n lhe
negr_iii:rtions whicjr prei:ed* :.,s i:11.,.a:i ,Ti{}vcs jrr;t i;r¡i:,;,r
'ti;¡otiaiion ganr*'ü * i.*" fr-;fi,'ralize ihe nerii-rriii.i,ir ur.oredure. ¿ !rrlrfl;1,,
,,, \ ii¡r' ilie Ío¡¡nei .ea.nte ,,\' i. ¡ ..,tatemr.nt o1 ltur,. t'-, a,,r,l,r*, tir* neil,.tri¿,tlr:lli
n0n-cooPerat¡ve cooPeral i't e
.i 'ihis ccnsiderátiün iail rnültc¡ eiloilnousll, lr¡;il:--k:: t,LltcIii1r.r)i- isee Bttlit:rirc igF-1.
/r\ ,/"t
/'. \. Rir'ti¡i:sti-.in and \!'olin:;kv jg.¡^"i.::rcj
ljinilo¡t- and l{el;::.r: ]9g5.)
/1",
/'l\
/tt z'
*
ú \l'i: ha'e in ¡lind piobleris ci eqniiitrrirrrn selectic:: rnC tl¡c Liil'jÍ;riiti::; r¡.ii:.i rl,i-
iri rrtr::tsivt',forll1 g¿1l.t1es or i¡l ga¡r-,*l of Ílc¡riipiete ini-c¡:::lti¡:¡n.
? I:i)r sl*lplicily, lre leave asirle lhe r¡uestion ol rtr:-;l;r¡ i:nt,:racrion Lrltliüufll ilri'rr
aon tgsls without side wittl s¡de
*',il bc ;ir¡¡ri situations n.íiure this is r¡erv tiunif.icant.
paYmeñts ilayments $
itecit games,1 .1,i
s
s
$
n'. t i n fr <; ¡l ti * fir:
,\,*s& lrcryateimg íheo'ry^: sn iñtr'Jdatri*l
i\ u :, h h s rF ai ¡'¡ ít?8: ¡,; i é:# * t t¿
'iiris r¡'¡fo¡rnetic¡l wiil i*$iai:r the same ev*¡¡ if ¡:¡:¡e *i' our chess-piav,ir"i
ugd*r all p{}sstbi* e,;eriiuaiiiies ¿ri¿1 i¡ov¡ f:r¡aily to cl¡i¡ose a *rateg-v tcr G
ciri¡r¡:¡an¿r:r; ir r*piai:ed i;y Capablanea provicie,it orrly rhat brc;rh I:ave th¡
üoriilngeni cn th* er¡¡¡rs* the negotiatio*s lac,k. 'fhe neFotiation gal?ts it'
is
sarne pr*1eren.es over ¡he sei of outcomes. A tr*hasii:¡:rai anaivsis rvsuirj be
f.llen analysed a5 * i:{,r¡itsst The sshiticn i}f ,,\i as a c$¡]t€st i}:en pr,:vldes a
üuite a. {.¡iiierent ¡ni{tter i¡nd w,ti;id doubtless iake a*cou¡il *l such f actcis
'c$l'{ttion ¡Joncepi' tur tir* *riglnal ganre G. },iash ¡naintalns iiia! s1l';¡':i$ii'Jn
ag 'cil*ss playirrg ílowef' elc. A¡r ethii:ai aaalysis r.voulci similaily requrr*
ü{)noepts forGshouitii:erega.rdedthisiva,v-i.e.astireloiutio:isr-ri¿ssl:*iai*ii
taking inro eccüiint iiurnei*us issues which ar* irrelcvant ro a galne-ih*oretic
n*gotiatiL, ;r contests.
;:nal3'sir. This is n*t io say that a behaviou¡al o¡ ar¡ ethleal anaiysis m$irt
An cl.rious r:-ojeciion 1* t-h!s ii¡le of aitack iies. af course, i[i the difliculty nct he ¡ncre r¡sef*i than a game-theoretie analysis: anly that it is as rvell nct
ol findin a niodel fcr tiie rreg$iialicii pr*tsúuíe which is botir realistlc and t* be c+llused ai¡¡uf what sürt of zuralysis is actuaiiy being atternpterJ,.e
suf'Ílcienlij'simple it: be afile&abie ta a:reiysis' Ürie response is that diff!- Thcse irieas {:¿íry úver to bargaining garnes. {f the ga:ne'tlieoieiic o¡:iin:al
cultíes |:nnof successfully be cvaded by pret*nding iil*i itrrey do noi exist' stretegy i*v*j,.'*s playing like Attila the Hun, then this."l.il1 rernaln ti-ue::ü
|,{61rysvr.r i.natters Bíü nci so liopeless as thel'rnay seer:! a: firsi sight'
!r¡ *r:¡r'
fiatirr w.ti* t?:e piaver: might be. It may be true that Si Fra;:i:is olAssisirnight
iideilrE eppropriatr: iieg*iiatio;i rnodels, it is neeess*ry' to ii¿cs Ú]i lliÜsc'
f"ind this rtratfgir harC to imple:i:ent but tl'¡is is equciiy t¡rie *i a chimpanzee
, i(its r,.i.iicir have, ür a¡]ilear tc !'ia.ve, I g?ll$in-e strategic releva¡lc* tire i*
atternpting the Sicilia* Defen*e Alternatively, siie may'. cbj*ct that St Frar;cls
,aiiclrr. f'his certaini;r d¡:es not apply tü ihe bulk sf menc*uv¡e! t:*I!lel{rt wor¡ld not r,",*-¡?i tü behave in this way on ti:t gi*a:irls that ;ire "*nrl does nr-.i
i* ¡eel-liÍ-* ne.4*:"i.:itions. l.j¡rder rhi"q 1e-rter ireading, fbr example, ei:lrn* {kii:ri- , jurlllv ¡"he ;¡-¡':¡";:s'. Bril thi* is an etiricel quesi:,:n ¡a be de*l:',r:iii¡, ii'at il!.
airuse, tiie incs*s:i::¿nl of bcrefit¡:r¡ at|d oiher m*re sr¿biie ai:-ernpls
J.';'} Ii"il
i-'y l:ler:a¡ ihe p-.:i .'í-i.,.'rl
iire opponent at e pnyehalogicatr di:aivantage. Thesil i-:.tilis rl.'.)ül.i t'f Itaií}iy
1Ve have bea"n *rguir:g tnar lhe itiasli ¡:rc6ran Í,;¡ f.ir¡naiizing iiegotio.ricli
'¡e of thc gieá¡e:r,i iniilüitaflee ili. a behni.'i*iliai a¡41:tsi$ hr:t lhe1" !i:vü :rcr
prccedures ii iess lbrrr:iriabtre iken it at fi¡¡t ap¡)ears pravirled that ¡:ne aCher*s
plaee ii-t ,r ga;n,r-1h*or*tie analysis. i. ¡ati*¡1*l player'.viil sifii.'ly igti*r* such
tirmly ¡s a garr!.-:h*or*tie viev*print. N*vereiteiess, !i re¡nai::s tie case ihai.
ir: i*levanlie:.ñ
wh;-:r the negoiiai';rn pi*ceJ.:-:ie h¿s heen firrrr¡aiizeii, r)rle rnilst rxpeú1 t* b*
It is as r¡ell t+ rnlirr¿e upon ti:is pcint since it is i¡:i issue r:it wirlth lhe faced rvilh a :lru.t'"xre whos.e a:iaivsis rnay be t*cir:":!eaily iiiificult.Fcr this
lite¡at;-¡re has a te$de*c:,'lc f't¡d8'i' {See ail* St*tion 'l *i tile inirod"*cli¡.;n
'| 1
.:i re*:i¡l ar¡d oih.r-rs,, Fias,'.i rvas very far frcm t',rgg*sting {het co$perativ..e g:une
t* ¡i!¡r üoí$p¿nlcn voit:¡::e') C*irsider, for exat:pir:, a twc-Fer$on cccp*rativ* r'
.i.iÉ theary -.r:r:';id bc ab*¡r{io:led. To dc so would be iike atientpting to st¡:iiy ti:¿r
garue.V+?}}.üeu¡xennafids{cigsnstein{i94.*}*riginalll'st}.g€lested.thatt}.ie i
is,¡lutic¡r-, cf sr-¡,:h r gam* be ir1*::tiiieri wirh a rvlii:ie ci¡;'lti¡iuum ol possibie niox,emeilr sf bilile¡d l.ralls using oaly <¡uantun: th-ecry. f,]n the cor:tie.v, ldash
'Í llic'ughl ,;i csoi:*rati;e and non-ccoperative tliecry as coaiplernei:tar-1' ar:d
outc$it.!es raii**i ri:i¡:':-!egc!!ati,-rn tet' ill'cr¡¡*' r:l'the garu*. T* ide*tiíi: :r
',.I
,t
mutu*113r *i*pp*rti:rg, prongs of a pi*eer ettack ilpün the prohlern clnegr,iraied
particiilai üurdofne *i¡¡¡i¡i i-hc {{:r*. iix€ir sugsfsied, is *n **:*rpiise *r:isii:le ,,4
s¿ilies. lhis *re.,v ls s,;netimes el¡$sitisd along rvith Fleg+iia* di*!,:c:ilc as i,ein¡;
,ropr* oi tii* gar:.i* :hec;i$t *n thfi gi*u::'*s that the r"j:i¡t'c''rE1e *ii w-|riei¡
:.4
tlr, il: inpe*eilubly parrdi:x:*al, b:¡t rqe s*s iitíl¿ jusiifi*ati*n i'or such an attii*iie
tire playrr-t :iilaiX',* agre* ru.:iii d:pe:id o¡r tireir reiative '"irargair:ing 5ri:weri'
i. lr
princrpaii,v *l:
certainl,v Io tr¡.,* ti-rlt the fliii**rT:e ,uí ih* ga:::e w!1i depend "" i:
I Thr: res.sclt l'lr att$¡npain{ ir ¡¡.::ne-lh*creiie a:raiysis ralh,¿r th*n a behavic;::a.l
¡.sycieoiogrcel facl*rs. But ihes* i*cis *re
i¡rclevani to ¿t galÍc tirec:i-st w\i:' ..:t
..',.
i.
it ¡;*;m:;:p¡¡:r.ir¡le tü Ecl?s ihe si.mpie pr*blenrs lrcf*re seeking tc attaqk
:¡:;:,ij's;r is ihal
*;;t*"s i:'
is"ecneein*d wi¡h wh¿t p''p¡li¡J be i¡e-c: lbr *:e pl*;rers if i;r"rli' th*y ;l !a
ll:
':i.i
.,r,ig iii !* a i',:::n;:J g*.rre as defici*d bl, volr lüeun:e:rn and S{+rgens!e¡:¡, the *nds arr i:,rti:
.rstcüínf ci ú abiilln*¡j and :¡*t r::lth lhe i:É&l arairle i¡ocr tlre rne¿ns in that Éarh t*r¡ni¡al :nod* catl 'be re*,ch*d fy *ni,v ore palh
Ldssr.rir;lng hls "-+¡:t¿li': wat trtli' '¿"itll the i:i¡ S
ii', rh:r.rugh the game iree.
rnafiliei i¡¡ ¿hich it was achieved' $
:r:,.Í
r:' ii
11H
5 ;\'r:sl; óoig,;ii:;:tg títtor.t littt'túu( ti "t ¡
rVasi: br:,'g*iir i * g ! h c o n' : t] t! itt t h) J Lr ( r ¡() t t
" '¡'t ,
¡,Lri*rtc,¡s ii ih* l¡lini 0xir'r1;rl€ñI1t lrn.l ii i:; ¡iic irie a oi r miriii c:lp¿:¡ir¡*rii
:;
guclsh'r'Íi'
,ji:i:,,:,i;¿| an¿j ihus 1o t¡iininii¡+ ilis depentlence <ln spcil.:1rf tlJ:l at;d
'1.
ti r',h¡cit is ¿i tiic ir$ált,-rf :i¡e ncn-cixrperative siiir r:i ilrr. hitsir prograrn.
i¡l '.,.' i¡i a!'t.3rFaii1,iitg is cr-'nce¡rtcd, ts"c i¡rl¡iledi:ric tssrrt; rr!:e * 'l'ire t:Llr¡irrs* ili ccnsi,i.ie t,;ri treg()tlaii,;:r r r)ril*sis in
to be irrlrrlec!iately inlcresii:d in the ri*tail¡ t which the ilrtaili ill
rii-iit* n-;odeller is urilihel,r,
per \Lrlrat he wiii care aboltt 1s hr.r-t''' ih* r¿:s¡¿J¡ "i ihe
9.
the r*got!aiio:.i procedure !lír spelied oLrt *xpliciily' and wititirui cr::¡.,r.'r,
rt)I ille fl.gOli;iioris se. ?
is ¡i¡r,- besauic ii i"c tiloughi that sii,;:ir tr)odels r.,ili trir¡rlacr-'tire use cÍ aj.r¡i:'r-l-
5i
liii: n.gLlriariclt pio,;rss i¡¡ f.ite ;rea io be c33lt witli !nf.rrmaii'y bv inti:11'ier: i¡
:lt
¡c ¡r:iirir: in,-r';lnplete ti:ccries.la The hallmarh l,r;d ilte es-re¡ice of tlt': i{usl¡
s:,,:iüi;iian cni ri:.i.r¡ns:1 gue;:!r,r.;ii-. L-r pr*crie al t*n¡s. this nieans ciii-''.rsi¡'¿ ü
+
Í.tralTt' t!1 i:; ti:ertfore íne irn¡ter*li.'t)e to fest sbsírdcí or itijonis! r<'t¡sat¡¡¡i!
lr: ap¡;rr:pri:1¡ '1',¡'-¡¡ereti';e ;oi.uiion cünce !rI' uoder \vhich heading carr b* };
r! ,,,,'i:h, ti;tt;iii íur :;¡:';t!fic negoÍlr¿tiúy¡ ¡t'¿od"':Ís. ,{nd an irrpcr'1ani clul', ir
ii',;iuied sr.tcir qr:tii-rÍs ts lx,ialrasian '-'c¡uilibrieni" ili'-' Sira¡rle.v vaiue" ihs
L:¡)i'e f,
¡¡ rl-lriie$ theriri:'is is t,¡ pri:".'l¡ls a batterl, of su!table lnodels lrr facliitate suqii
Lri 3iicl 1"],:ilil3rr¡r s¡d fi!6rgelstern soiution sci. Wiiililr iire liesli prog¡ii¡l-!. isl:rg.
ci tlies;, a;'rC $tilci co,.,.:.re!'etive s,:lUtit¡¡1 crJilcaills aie seert a' ail atti'illpi
$ i
e;cir
As an exanlpie, consirier the r;se tiraf De h{enil (1971} and othe¡s have
to rjescribe tllr {iú'riro1"!ie'¡f a ¡ron-cooperaili"i ne¡l'iialion g'tt'te *'ilitrtttf
s
course' a coope rative rÉ ¡:raiieol tire liash bargaining soluiic;n ir, discussing unioii po\1'or and i*ialed
;i:..r:;ai;;it,{ !lic negr:tir;tit)r^t procer}urt:. To be useiui- +f :!
appliiable) to
g issues. Froln tlir p¡,i¡¡ *i,.'i¡w i:i the l.iasii Fiügrü*i, suüh !roil,, il:"rs¡r*r:1
:-:i.¡iitirr.i¡i irroileilufú rlu¡¡ br'ap¡llieable {cr thr;ir5ht to L"e e
$ ir;lai¡se it Í,J?:; i¡r1 s¿ri¡,-,i;¡ir.,¡CCress the questr*n oiirit;,r, the iiish barta,:rr-lp
ialii' v¿id.; cirsl ci leg+iiaticn proct:dules. in s+tllt: ia:es. sucii aÍ; iil* sh3¡]if v ::
q
r¡iiitiori ir tq¡ i:r' us*C ¡¡iher:iran sc¡ne cther c¡oüeraiive iioiic;-t. Lik* rearj,..
.;,rl',re, tiii: ColtCapi is i:*st see n aS an 'aver:gr' of lo;li¿i i:; to'bl expected o"'er $
as tht c{ifr of a
.x
;:,;::.'rc liill:,. :i,r---ii¡.',t'-.:t i{-}:-jr-rfi}'i5 giff:uiC nOt iif i:oilS.i11 vriti:Úi,ii ilrst t¡',.ii-it I,itcnl
sUCh 3 Class cl negotiaiion prücedurils. Iti otli.'r cases. lucir
.¡
cortccpt does nert suppi3';r Ii; ;ii'tr.;i sl¡i:- A:t.,i ire¡;:eri:. r:i:ent rvo:!:. based on an inrportall pap*r i;r'
",,lii l{eu:can:¡ ::nrl Mo¡Ée,nri*ril Soluticn set, tlte :i:'l'i::::lcin i,:hriFt*r j ,,; j-lig alrirciri r'.;lurne.!, indicates th¿il sr-¡,;h i.x-;¡:L,r rr
liC.¡"rniltaci ir: illl i¡l'tili'¡h !
i:riiq'.;e +.,j,..;i):lii and *-,¡r. is ire* lo att¡lbui* il-rii ;
intcrlcitLl i$ he *¡ririi;able' i: , ii;s"i!i*,1. ihr iirifl;':lt_r' corii*iit$ '*uiside option-c'- b;,.' ¡i¿¡ rvt ntrarl r.1
*i iie eiass *i'ilrJcedrj:.i:,.; ',r,hiti'i iile cDnCr:Ft is I
,i¡e iriliiii, ir-'v¡:i: whieh harg':intrr i¿ri auhitve i:¡":le*.'.'irlg the b¿¡x:.rti',i'
Tllese lssl re¡::aikl-r:n¿!';-i¡ *:ital ti:;t rlrt N¿sh "'it'¡poinl i-" vrliiabl*
as ¡¿ .;:
s :,rhli: ¡l¡lrj ai.-:ir¡d¡ning tfte n*gotialio¡rs. 'l¡aditi,]nal rvi:Ccin takes ibr gralited.
pir!ic;.toph:;a1lriopi:rtiiLlx'illgahcuti-'ilcP*r¿ltivilit-'i*iicfii{incepir'But¡ilrr l.ll
*,
!. ri;:ii lilllr shr;'Liiil bc ¡ica1i witli Lrf icie:riifying ih+ "-ctaius qlio' point \1 r:i'
rtal i,¡iis: (.;i'ili* t{arh or,,g'*rn iies ill tile poirrt to i;;:':r:i1* ne'}il. ihe fi ,l; "'.,,¡:1:rric cpilon'pcirii. Ho'-','ever. rii'l exarni¡i*tio¡i cf the apiiropiiir-
pi;1,s1r;*1 sc;ences, exp*rimeili prr-r.riies r f inai couri ;o
rviiicil criti''s r)i '\ J:'
gU*ssw¿:rk '-:an rpt:eai' ln t-he so¡''i;lisCitnccs" ct-tiiciu:;i:
ts
f .:t:l:¡il¡l of iiiil:ii¡isieirl r vpr\ 1,,,¡y'¡i. ¡9¡1,¡¡1,rpi-,rati','e bai6ainirri¡ ;ir_rii. j
ciSe: in s,-:tui-lierl.,rn ar.r.ei
.ir,.,,:: tl.i:n. ,¡'lriit: 'ri:i: iriertilicati¡;r -r;.,-v liave r**rii r¡: sorrr* :i:ecial ci:-i"i;ii
*
a¡ips;is rc trpei.iri.:errtai ¡'iidetcr: ate ssliüu: rossii;i* pa:il¡' bec:'¡:': "'' *
{
'rj.{ü¡"!ie it is leloni fe¡sibie i* impose tire ¡¡*'-rr'l' ñ i,aili.'rs. it ¡l ¡¿ci i.ii*: rigirt !1'ai,' t* pr,ieerl in ge;rtrel,rs Tire pairri ts lh;,1
n,t,.rrirli;¡'s !-,awit ¿nd IJartii se
É
!t eil ihe mcre in-rpr-ri-taliÍ jn tiie s*cial scirler' t"'
-qa¡y ccntrcls. l'l-iis makes :+l l.l i'i;i-r ,'':i;rt is r¡ailt ¡i;:re;rer !,:r:gil: i¡ i-!rr i;:t:¡ducticr'' io ihe cc;:lta¡icn'.'¡lil¡¡¡i:.
afe íiot
e xercise r¿.:rain1 in rn:iking assuinptions * pá.114üiaii:\i llrr-¡:¡t wiiicii &
É i /l¡S*¡¡!:¿::i*ns cs üc¡fies.
t 'tlitra?¿¡r'
i;f r¡¡li,:i b*i i¡::lii:c!tl'"
statL:rl e,:tlictilv in ihe dls,:rlptlcir 1hr lorrnai ':p¡reai * l-: r-ir. i¡¡l üacilsio¡t, tlr 4g9;tt ¡n the t-c¡r¡ul¡tíoq ,-,{-n(:w a*üFet?iii'/e soiufion conr'epls.
..i il¡' tirc 1',pe oi p¡r.css, iong far:riiiar l¿ ir;aih¡:¡i:;{ir;i¡¡rs. d¿s.:¡ibcd i:¡ l-akat,.-.r"
ou , ,.agri, r,¡i'tir¿ ci,.:ic* ,¡f i¡, stiuttu¡e tt'¡r the fori¡a1
tnr-tt!ei anC wh¡s" -:'¡stl- +:..
r rs,rrt;ol:- t¿t1ti. g. (, ri i ú t k) | s i 1 9 1 6
*
fl:
üry theory: an intraductipn
Nash bargaining theory: un introduction i I
ti j ,tuusfu bargcen
fti-.q1.-^tll?i:
r¡ldit io nal *,r¡c,c:n" rvhicli
is intuitively quite compell.inU on'
i
without tire üiterventto¡t 3.1 Commitrnent
\\'r)uid nLrt ;;;;';;',l chniienged, let aione altererl'
advocated by tlie l{ash program' ln niuch economic tireory it is taken for granted that agents can wrile what-
!)i a lesting J;ne **p*ri*ent
as
tli'er contlacts they choose and that tl¡ese contracts will be totaily binding
on tire players - i.e. that the basic liypothesis for a cooperative garce is satis-
rRc}cESSns flied. Two defences of the hypothesis are usually advanced. The first citr:s
3 Tt{s FOR-\{ALÍZATION OF NEGO"tlATi0}i
the existenr:e of a legal system and tl¡e second refers io the importanr;e of
cruciai
dlfficrril :* n* to identify the inaintakiing a reputation for good faith in so f'ar as future transactions a¡*
At first sight. ir is surprising how 11 very consiclerable
rea"tuies *f a i.,a¡gainin,
pr"J"". o,'e wouid t]únk that tlte concerned. kr ma.n1' cases both conside¡ations will be relevant. Whatervei the
p'*Ut** vi'hi¿il lras piague'J bargai*ing irriitilaterai)t ttrlaf the pl*ve"rs may rvish to make abr:ut how ilrey prr.,ci:r
cerrsr íor tir* ia¿i'te'm'i'a*v 4' 5 anc 8'}.
the s"rblect ' {See chapters t,-r ;olidu!:t 'Lhr negotiati¡;ns in the i'riiu¡e. S*iieli¡ng {1?60} has emphas:;r*d
since F-i-igewortir ;;;;;; is now
"-rer-Ihere prucesses'lut'il brrt"li hoir¿ useful it is to ple;,ers i{" the i' r*ii lo¡v,r lheir c¡:porre nt r'.i!h a
is stili much t' üt t*u'nt¿ aboi:t bargaining
urr! t*"o0 "*uiri ;le:tiii¡:e "ial<e it ci¡'leave ii'¡:rcbier:r an',1, ein-rr"iitalteeiusly. hor.r,hr¿rrl ii is in
ro1+r to be asr;rib¡"'il :i
:::*":::l:t:.t:::;li::
'""- . ",
iairil' weiitsrablished'in'i
iairiv weiirsiaDilsneu
*nui tt likel¡' to happen
"*
Lrral
without 3 on
l1¿r7i¡19 v10\ru 1"he
io
prr:i:ricai ¡etr!ri tcr tnai.;r lhe necrsii:r-v cl'li:;r-iit*iit:]i.:; stiik. Pelhaps tirc b¿s:
rin*u,r, exarnple. altir'ugh in a:L:,;lre'¿'lult tiii'ftren'¡ 0cfltext. ls that of Elisbeig
a comrnitntertt: tl9li). l-ie cites tiie l.iil¡i;rf:lpin¡, vicri;n rr'hlr +¡qlulij {le;lri;r 1yisl1 r* ntak* ;:
h Í iilie ; ir{,.1rilriliimeiii not to ¡erveal ihe iclen'¡iiV of 11:* i<iCnapprr in order io eic3l,.*
c in io rr¡ at lo¡r t''ritü.(I htrt ha.; flo w.r) ro ur) \r'.
shali take these one i]t a time' Oi course, i1. does not lbllow tirat, i"¡er:e*ss lh,l kgal !y$tefi is r¡itiliirj,,'
withur the bargaining process' We ; to be ei'iective in enfr-rrcing in-bnrgailing ccimmitrncniri, lhat sitch (".,)r,1,,,;t
oi¡ tJ¡er*atons' 'fhe simirlesl ¡{
particulaltrv profitable' tÜ :pr'rclliete ttlents tnaV flot be enforr":*d h¿,caLtse oi th* C¿rmage thal a lailu¡r- ii) il;:jir'1'
ie t, ,, ear¡ ' if not
psychclt*ü':*u"' dominate u"'-"i*"t- ol rational issües {as in ü
rrplanation ir tirat "t equipp*d us t'-'ell rcr $ thr';u¡lr cn a cornrnitinent rr.lay incur in respeft ,-ri ;: ¡:1ayer's Ífpütai:*rn.
social euoittticr''i'it
p*ktr games u*to"' nout'-J;;t"0"' ""t
si¡ce,I:"::-;tl"::::-J":1;;.?-";" -"" * Obvruusiy, a reputat!on for 'tougl'ines:;'can i:e vcrir valueble ii i:r,ir¡li.irir,r
l::ilr-l,;il.',""''l'" *Jtions ut
ricaiuig ivitl, i:aisaining srtuations ' o,''":1?:.:T::'l:,::?*:::.-¿:lnt;$::;:;:1
isoration
u uuri.ty or ¿ii¡*r,.l"iirtues
Fnssibil'thepr'rbiirrr I rviih ar individual lvitii a reputation fr--r¡'softnesl','f'hii is anolhrr r¡airfli
our bargainirg parlnfjf$ * ,n.Trirr."rn we sitaii lake rip again under the headirrg ci'inior;i*firn'. F-or trhe lr¡r¡n:*n¡.
[::{:'",*11:ffi ii5Tr;i,*;;*ff *:*:'":::;u;l:;:r;::'j'ii1t*ftir[i
have a strong t*trtleuc-v to
p'**na that rve havenotbeenbargain-I i 7 Onc could, of course, reÉard the players al bargainirg ovr.¡ their valuatiL:irs oi thrl
iilncern*d, e'd cortiinl)
rng at ali. A Inore *nto"'*glt*
hyiothesis " *^tt*itili'in' l' 'f'ut our' hargaining$
F':ssible qoiltracts or restrict aitention ta "self-entorcing'.it:r1.racis in uhrci iJo,.]C5 ¡ie
o, *nii'l'"'ii '*'
i'"r|v"*,l). e:,changeci a little at a time (as i¡ illicit drug dealing)"
events L:"if ::tr:TT;Jil""*:r::;t:::;
i,*i,,¡uior,, i:; rhapcri \Á'ntiln aic ¡rul ¡lvar!¡+'"r
*"1,iun",ion ,*qui'"u taking account-of $
18 Or, less obviously, the Westinghouse contracts o;r uraniti*r supplr'.
r',r unli.k'el¡ io be properiy understoo<tr ;
ú¿a¿ inirlg theory: an inio¿lucti* Nash botsainhc theory: an iñnoduction 13
l: Jv¿¡rr il
,o ¡esis,er thát *-"i,lr""ghr.
of v ¡o
wi3[ oruv assumptions::y":;::':1fr:i
"''"'' "-' rh_ey n :::;:1,;J:,:"i::":'*.",'i:i::::":::.::,?T
hoscle'¡ we wish be l::.T:1.::T:r.'1.-"
however. t¡ so far as . f the plavüs js ¡b¡e ¡o wrn rhis race. then we w¡
wrl¡our.care'* tett wfh ¡ sracketberg
-"i,,,",1^j'1...,,
made wnnuv'!
D' m¡qe '"'ln"*",fi"r
_l-li,--,,""¿*,
'.nrnnrmenr should nor bc
cJ,unnmenr
Ie sLr.trindble by appeals lo iepulction 'deas { rszs.
Lr"¿su 197c.
-o {t{ readetsiolosei game. A more
reade¡-iolloser'eajne. ¡,.*.¡"" siluÁrion wr¡ srise
mo,F irterering ^-,^_ -.
uhen rhe
ffi:'Hil'Ji ffi';:'i,'i,'"i.iiil.'
i'ens dnd wilson reóz'
ir
wrren
envrrorunenr ** "^'"
''"" **
.L-.
-i;.;
r'.""inirp rete
' r""""""q::::,*T:';il::üi:
'" '- *rr''"*-,erm 'L-
consrderarions
toni-rerm considerarion" rI
prrceiranen"ironmentúi,Tf:j.1'.I:":,".",ónore,m.onsideral,ons
^i¡.e
I | iomnltmenr race r! fior winn,hlF in,h,,
.!l-!",lr.o**o s,muha¡eoust'. ln *¡'"..
ments "n
"".s -,",,^-,"
;:iliiTffi:T;,T'':J.fflil',1j1'j:"*.,i1:':j'.:iT-:'.t.::i' ^----, -
"*. these opening s,atc.
¡nen,sm¡shrconsisror¡comprex.¡re,;";;;;;;;;;,""",,,,"-",",",._"
m'8ht consisr of¡ complex, inrerrocr<tngsysremorcondiiionalctauss
dirñcuxy
:*d,hre. Bu( ,h¿n ooe has rhe
to 3uch an exrent'n1'
"l',\:;;?";; ;;*"**br. ," {
--:r'¡¡¡urihrv
¡o, d,e un,ávelrenr ,:"¡r.ii * ,.,"v
* r€quired. rhe
¿,e l,lel! ro'¡nlrude
l¡kel) ro'inrrude T"'.-:'.;:,:,,l,;",hpv obrarn I
úll oblarn
rhey wrll ,uie t'rthe
ruie "r
lasyer" would be
rhe rasver! r.,rrh "il,.i;:;';;lo
bÉ ,. s€¿rch r..,
torsj;int
$,ioo{ r}ir rhe praven are morivared onrv b} "- -" ' il
"É "Í *r.''"r, 'o coure ofacrion forrhe
hánd I ¿'y"" *.,ia ;;l;';;;;;;í;'.;.",','": ," essence, eich
¡s" resurr ol rhe deal cu¡¡enrlv in ¡iinculrie! ,s 1o jl
oh\iotrslv lne oes¡ wdy ¡u v!6¡
il:ii3i'11."ilJ";il'l"u'*"'*i.n
,hxnooroo¡ üe ar."mpl
altempl 10 mudel the
these conc¿p:::i.:':::1i:::,,::"f,
impact ot repuraticn considerrri"ns
. r1!^ r
ra\out
I
t
:ii',iT,{
trn¿lse
rn¿lvse
i pr,y., woutd be announcins;,".;"..; ;,,"",,;-;-,^',"^ ^,_.--_ -.,
ii:'1::1f1.'i::;:
o'her e'3vers i:r'"J:T:iTi, l:- il':,'n.':"
and pe would end up wi,h a .iorlowe,-toüoser.
..rir'-
carne ro
'n.' re ¡aüsinlns s¿nes In I
rl,roüg.t hvpothesit abour comñitmert ': -T:."d *, $ $tuch lhe piayers ll h nor going to be ea\y ro sa] anyrhing very dcfi¡¡rive abour
.'r¿moder ¡ which conün hent " "' .""'li:';;*;'b;-r","r"e d* in ra''er siruarion reyona rre ,i.i ,iii.,i,,ry,s ritery to br
pl-,, ,hei, nesol i¿,ion s¡E"st:' "l' "3-v":l^l-,-" {1 'nr "¡,1-"¡",i
.u,up,,s(d inro rh€ openrns jnsran¡ ¡"-,r,", *,^, "^..,-.,^,":
Íil,lT:,::;::fi:T-lT::*;:1";;'";,'o*'*.*
rririch theY are involveu ",-FR in
{
which rhey m¿y
*T.:i':l:i'1.;l;:,ffi :::'m-:Í,J;[ 1ilffJ];',,'i'#i';i
ot the
l';1il-:l::-"r:':*:::::
_
b! 1n'rthei p¡'l:1¿1'
suc}r Commitn¡*nt battles
in s less abslta;t LLr"rtc';t {clrrscJ'
(i96{-), 1966) for I
19 S¡e sclielling
l#
írtt.roduüion
Nastl batg*iníNg theory: un introductíon l5
ficrh ing fueary'" an
14 lrcrgcin
is often solved i
1g
ln assigning the entire rnotivatisrn fcrr bargaining behavicur to the efiects
problcrn' ln re¿l-lifc' tlus the t of the passage of ti¡ne" perhaps cross nverstates tl-re cane. ln the absence of
lre taced witlr a coordinatioH
g;unc
,\v us*r* ír1 essenriattv ffiffi;';'un*n t*'-lrtá''- t't:l"'lt]Ll on a cosi- ir¡lormationsl proLrlerns, however, it does seem natural tc ftrcus or nodeis
( i')6U) giues nran¡". instru:t:l:
i*umplt'' Llbvi'';t'sly' ;
ir u'hich commitÍ¡-rents can only be made to sticlt for sl.rc¡rt time peric<ls. ]n
i rsht'). Schelitng ton*i'r*ntiy'-!]l:Ottt i
unless.it *plo:ltf 't l sueir models, a player rnay threaten to delay agreernent fperhaps ¡or rver).
rrnrtl',n i¡ '¡i iro ¿reat'use '' ton'i'i*n'y r¡f the conventir¡n u-qed
t:f requinng But such a threat must be credible if it is not tg be clisn¡issed as idle bomtrast.
expkrres the conseqiiences sames' *
*ii'r" "oii'kil;;- demand Ir is the'efore na.r-u¡a! to follow stahl (19i2) in framing a non-cüopcrativs
to res<¡lve tir* coo'dinat';;;";;;-" must cosreiinate play r:n the s-1'¡¡¿- ]t
cf such a r¡.r*del in terms of the nation of a sr_rbgarne-perfett *quili
lt ls argued ,tr*, u *onui*ent convention before'
enalvsis
briu¡n. (See Section 4 and Seiten ig?5")
¡nef¡'¡c lrlasi", lrrgul*r*6
,oiution as
."rhre the clenranr
llash clenran'd
rh¿ Nash ganre '!
l"*"o'uit'-"Qrnrnitment is possible' t $tah.l (1972) looked at finite-horizon models in v¡hich lwo players alter-
Given lhal probicrn in baigaur- i
urttill.ltterminacv nate in making proposals
pr.'viries a sairsfvttre "t:";';;';;;it'^* il-t* e\¡¡'nt to which i agreement is reached {i{'e,ier). In cirapter g.
it untt""io'f'uUu't a flnite&orizon model is 'ntil
in¡ iituatiurs' Bur "vhat if there o-tallen up in chapter
7 t stu.lied in lvhich the players rnake simuliaileous
propcsals at a succession of times until a pair of compatible pr*posals is
committnents t'* *t*t'n*'ol"t*'^O'*t"nftn'J"i,,'
;¡-i
by v in cen t c' u'ut'"¿ "' ilJ';f :i:* lJ Tll;:J lfi1i::
T¡::.:,::::,?:' l *rade (if ever)- l{owever, the most signi{icant (and ceiteinry the most elegant)
at all. I{ubinstein. A very wi.le class oi variants c1'his m*ele} vi,:H the.\:d,¡/rrjton
:iusir:'r {alt}:lerugh 6¡ ancl frx may br. tiii}'erent). cirapter j examines su,rne ol'
F'oldts (196¿li a¡ h*::r'{ree also Macl-ennan l!}8ü1. what we there}oie havc ís cvid*nc¡ i,-rl.
relateql wr¡rk erl'
Cross's "¡olk and 'iht:' 'rhe ot}otatiorr $lliclr n' 1'rvi rs lepr h.'valu*
21 irt'¡ ¡n ap¡rreriatioit of ita:'i rrg::l \' ol an asvmmetrir: n-ash.llar¡¡*iiiing solution in certaift íilluauofts
tt;r.kr ,l r{¿! :rec ',t ',t''"1¿l'"* ono ["'o-'*ution in B::tlult;ing'
{ l':fi4
Jur.'ri t .:trr
r r'rf,r1r l'nrq *"*lti--""t1'l 'l'*"
Ndsh hargnin liig flreori'. an intrt¡dt¿ttiott l
l6 ,Iash bargawting tlteon': ¿7ü i'lrrtil'iLtc{tt)Íí
llle lesst¡n is t.iiat. wiilio¡-ii cornmitnlei:t, onc clilIi(]t af-Í'or¡,j:-. trE,.::ri,:,,
*ircn ti¡nsacticn ct}$is are negligible. It is tnr¡ tliat d:it* abor¡tr tlic mrni- ¡iri¡ui tit* insi,itutional assuln¡rlions rnacle. Itelns which seenr of ilttle ünpu.¡.
micr$ slÍut]Iu¡e of barg;lining i¡istitutions in usr: i:rc rinlikeiy tc¡ be srrifl- tünce at iirsl sight can sometiúres be ol ct-.nsidcrairle signifli!:irn;J ¡nd '.r i'
cientlv specific iú allou' r th.'orelical calculatit'n ol'$1 lnil fi:. But this docs repe:it our i¡bservatrions ol'Scctic¡n 2 that li¡tle enlpiricai uiirl$. tras bril
ilot nte3ll that ¡j1 end f: .nilnol lle estimated eilrpiricaill'frofir observrtions ,ione lvhich is crl ii-nmediate use to a gilme theorist on this issue. Liarlr',g
of other dt:als using the s¡¡re inst¡lr¡¡ions" chosen an úlstitutio,rai franrer¡"ork. an analysis is then nccessür\ which i:i
Th!.s has a bearing on w'liet has colrle to be called the'coase xhcoletn'
Lraseri on the tirne-dependent cr¡sts faced by the bargainers. Tiris raises ¡
(Coase, t',rirtl). t¡ cil.tptel 3
ise e also chapter -{ ), Rubinstejn s}towr
tirat
nun¡ber of spectres. Tire first is th:e gene ral protrlein of rncrdt:llüig i¡rlormailr-n
Parett.n-in.fsrior. subgar"ne'perfect equilibria exist ior a'Írxed-cosl meidel
wliicli we treat brieflv 1-.ciolv. The otlrer protrlcius are tireoretic:rl proble*ls
in wfrich it costs nacir plnl,'er t:t (c2 0) to d,ll*y agreemeni b)'ilirle ¡. This in gan.re tlieory. Wh¡t i:i the c)orrect *quilibriuni idea to use i¡r:rn extensivt"
¡esult sun,i,,,es i,,,hen the lime interval bett'een successive bids b*cr:nles fon:r ganre withoui conrmitnrerrt f What is to be don* when there a¡e rnultilrie
negligible and so lrat¡s¿ction iüsts *ease io b* sig;nilicant' eqLriiibria'? These and othcr r:luestionr are tc)o large to Lr* treair'.i in mc,re
Tiris resu!r:rrtj. ¡t1orc'si¿nilicently. thai described in tii¿r previr,us para- tlian a üutrsor\¡ fashion in this int¡oduction {even if rve titr:ugirt rve co*iC
graph. shr.ri,il'rar tl.re irtrpact of tirne-clepenilent eosts on the strale¡-lic providc ader¡uate msw*rs). We shall, hclv;ever, make sqntc ct¡nl:rents iri
t rhuoir:-,u, ol Ftra-vers does n¿t necessarily b,:c,omi negligible as the cr:sts
Sc*tiari
themselves Lrlccme rregligible. Sorile care is therr:tbre necess¿txy in inter-
4 .
preting Ihe res*lÍs i¡ lnodi-.is in wliich such cosis are set equal to zero
a prlorl- 3.3 [nformatiou
Acfuall), the pcint is m¡re l'undamental. Suppose that the elemcnts ol
I-jntil {ai¡l}, recently. pieces on bargalning thecry had little or nothrng tü:,iri1'
tlie descriptron ,¡f a fbrinal ganle flre listed under tiie headings
about ilifo¡nlationai questions. Presurniibly this w¡s trecause I-larsenf i'"t ther:;i'
a strLiiluie of tlie ga-rne trte: oi'garn*s of irrcomplete infor:nation'needed Iirne to perc-"olater. Flonev:r. ,.
1,. preferenccs: is obvious rhat l¡rhat the ¡il;i-verr kllow clr do uot kn¿r,r'rriatleir r ¡rre'il tieai ul
practice. At tire s;ime time tllere is ihe prospect tilal the bargaining.qilivajeti:
¡ inlil¡¡::rtion.
of the '1og of war' rna)- servc tro sirrplify l-he c1a:s of cccrpr;'rai.ive scilutil:l
Tinre is ;t signliicant iai;io¡ in each ni these calegories. ln tlre quoi.rlro.'r concel-1trs th¡t rv* need tr¡ iaki' scric¡r:l','. f-iieariy the se conccp-.ts üaJlnot depen"l
r:,ffered abcvt. Crr:ss (i 96tll assig¡s a domi¡ating rolc to tire nlariner i"'r sli:iipl1, on parametrrs rvhicir iir* +nly knt wn i,rr)' tli.rÉ\roxima{elr'. Ignt rattil
which tit.it¿ alfucts the preftrrnces oi the plarers via their atiitudes to time- r1;r!' nol be as successÍi;l a lei'eiie¡ is death, but it Lai ttr uses rn iliii iilc'tic:".
{ep;nrlent ii;sts. Eve¡r il inioirnetionai q';es'lions are lelt aside. the¡e renrains Lln the *tirer hari.l , rh¡ int¡oclL¡ciitn oi inii¡rr,latlsrnal questiol,.r i:¡r¡ r-,i.ii]'
tlr.: f.,lc rh;lr 1i:¡lc pla..s ir tll.: rlcscriptiolr (); ill. P.sjlle tree represctrtittg 'li' r-:cmolicate i.hi: i¡rmei, no¡r-c¡r¡pe:ralir.e analysir of bargaining conlesls. As
inslitr:rional frarnework '¡.,irLüi which the ¡:lryers bargain - i.e. who cun iu f:r¡ as the Nrsli prrr*iirnr is concernc¡:1. ihis areans tl:al wc have tr:',rork mu;ir
rvh¿i wirqtr. Rulrinslein':l rl,r¡rk i.anrl that cl otliers)?2 nrakes it clear ill;it a hrrdeu l:u1 with tire hcpe of i¡ less diverse ciass af resulrs.
classiiic¡iion oÍ e ooptrati;e bargaining corli--f ptS musi ne cessárily be f6r:nrled llll o:r¡¡rtion may 1,.e iar:xirrg a'llout a varirt-v ol íacti¡rE in bar¡aining prob-
olt ;i.ir airprecia¡io¡r ol the iiiif*ier¡t varieties q¡f insiitutlonai framervc'rk leiris. 'I'irrs volume i:oiit+iu¡ chapteis whicli tr,;al uncertainty abol:l the extent
v¡iihin r,''hrcir bargaining takes place. One might say that, with perfect irlfcr- to rvhich commitne¡ll is rrc¡ssi'ble {chapter'i). uncertaintf ii!ri:rit i}re preÍ'er-
rnation. tl"L,J iii¡titutilrlal fralllervork detenlitres the qualitatit'e natu¡e of lhe erices thc olhsr plírvers.nia.,,'htvi, r¡l'er the ncssible rie¿ls that c¡¡¡ be reacircil
rrsult rvhile thc costi ritr¿¡mine the result quattritaÍbel.1'(even thcugir they (chapter 8) and [:ncertariri',. over time-dependeilt *r]sts (chapters I and'll.
rn31, b,-' srnall in an absclr-rle 3ensr" j. This w¡r'k depe'rds .rn H;r'siy.rj,i's theory of '¡Earnes *f'!*cogipletc iittcrn-l¡tt.:rt'
as cliscusseil in the coniparriolr volume. The iheory rieals rvitir cliliereni-cs :r:
22 Co¡isiiler" ic,i er:arnple. {ie rerult oi rern<.^,ing the :,pparentiy harnlless assumptlon
the inlti¡nation ihe plrver:; hav* by introducing ideai'chance mL*\'.js'1¡,'ltLrsrl
ot Shakr.C a!1J sui¡on i11 cilspter 6 ti *ris volun":e 1h:¡! a t'irm rnusl alrva,vs listen to:n ¡r-licofile is observed by st-rr'e ¡lia1'ers but not iry others. This all¡ws irid',r'rs
'insiie' r!ciker's cortnie¡c;lfe¡ befar* s$'itching to an 'outsider' Withcul tiris ver¡' reason- lo have infomratior: rvhi*h is s*cret from the other players. i{cri'evr:r. it is
lbtrt, assurnption. a 1\'al¡asían {-1¡icomc,"v11! always be obtained. see alsr-¡ tsinruorr: (1983)'
jili,r!c¡rr rrjr
rii¡.,-^.+,,-+ 1... r1..,
Lrrv o,,olysis
^-., tti' ;, 'garne of incorrlpiete infoi¡lraii.',n' Lliai this
$hakec anrl SuÍ1+n 11984¡, itubiü!tein and Wolinskl' (19ü4). llinüicre and Herraro structrrre of the chance inoves tlrernseives (i.e" what cati happen ancl q,iih
(1985), L;aie (1986).
ll. \rars& &crgc r*l in¡: i¡¡ i',:] ¡.tr - ú n i fi trr:" d u c t i (: ) t'l Nttsh basgaini.ng lltttsr.v: {ttt ittLr$(lLr€tiúr? lt)
i,Lxed pror:eilurs by rols.m Í{ere, as ¡rreviousli, the legal 'proüess seíyes as ii
what ¡,rrobahiiii!*s) rr¿ri be secr*t" Th;: theory lhcrefore reqrtires tiie existence
surrilgate lor a wrdtr' ¿.:lass *i í)rocedures including that via which reputations
of frn iiiiLirff¿ilottal pr:rol in the L:¿;Kgrcuntl which is üÜntmt)n knowiedge
1984). T'he basic idea is inay tle established. Fublic knorruierige is snytiling lvhir.:h may serve as an input
{Aunr:urn 1!¡ ii;. },{ertfn$ and Z3ínit lü83, b{versoil to ¡liis prccedure. ln general. public knowl*cig* will be a sub¡tt of conlnlon
reascnahly si;"nirl* but theor*¡ical diiíicr¡lties intrucle collcerilii1g hoiv hellef s
knorvledge in that publir: knowledge will bc c,-¡ntmon not only to the players
are io bqr updñi*d cln.s time ilniers ii';r pii:ture.
{,'ha¡rter I contalns an ariai}'si$ (ri }'Ji}sh's dernand ganre in an iücorrlpiete but to outsidsrs too.
infbnnation conif:{t {Sec'ilons 5, b s.iiri ?). Since comrnitment is assurned irr Any contract at which tire players anive alier bargaining must rely onf'
this analysis. llme il not a significs.iit factot and so tech¡rical prcblcÍn$ ar* cn data which are public knowledge for how else could it be enfcrced? It rnav
rlini¡.r.riz-ed" tri l¡ s|¡o*"n thilt a vetsil¡n of the asymmetric Nash L,argaining
be, for e.xampie" that the bargaining contest faced trry the players is one ol
;lclution inlrodu:*cl b5'F{arsan¡,'i tnc Seiten (1972) is relevant in il'ti¡: ;'';:r- complete information in that tl-ley face no dciubts a1.'oltt each oiitei's von
Irxt. Ciraptr:¡ ,,- inir*iXur-'es a tl:,'c-:tage rnodei. but dii'flculties wlth ii;r¡e- l{gur¡ann and i\r{orgeirstern utilitv funclirns. But it does not iollow frc¡¡n the
1'act that thi: ul.iliiv ft¡nctions ilre conlmon knowle dge arnong the pla.vers tliat
ri*:pcntl*nt beliefs er* not at issr¡*. The later part oi chapter E iSectio,r 5
r-¡r reQ) represents an eai'ili atti'mpt to grapple with these difficL¿i;ies
rviti-.cut they are puiriic knrrwledge. 'To cstablish such intbrmation in a court of law
th* l:ier theorelicsl fr¡,*rewolk r'-'hie L hecarne available - no1áilh/ the Kreps* {or ttie woricl at iarge) worild iequire an appeal to ther theory ol revealed
pref*r'encc.?s But then a piayer v,'ould have an inccntive to reveai fals* pref*r-
wil:;an {igs:ri noticn of a sequrnlial equiliilrium.'r}ris is em¡rloSred by
ence$?r' and. aithough the otirer players might be aware of the dere¡rlt'rt,
Xtubinstein in;hapter 9 wherE he *x.lenilsilisdisec¡untingurodeiol'chapter3
tothe caseinrrirr*htl'rereisincor,rpirteitforl¡ralt'rnovertheilislot-il:t frciors' they would be heiplers tú prcvent ils implement¡tion. Si¡irilar cr nsiderutl 'ns.
-i-hese
ihre* t.:hapt*;.: covei a feirl¡' i.",ide sllet:tIufn, bur ii will be appr*':iated of course, ap.'ply t,r ollier pieces of infc¡nnation ¡vhriclt lr.¡ay not be pubiic
knrlwiedge where the .playcrs have tire oir¡:oiturrity to distorl the avarlabie
that tile basic problem is tco large to l-re more than just tcuched uprln in
thls vi:.inlnc. lr;i:me¡r.rus artici+s23 l'lave been lvritten lvhich treat clifferent
er,!den ce.
Suspicion shotlld tirereiore be ciirecier.l at cooperaiive solution .ronr;eills
aspr:üts of the prrr'nier.it. bui 1{e are slill just feeling our wav in si: far
;rs the
o{ inc';rnplete inforrnafion is concer¡ieil' ivirich take íor grrntctl that cornnon know'ledge is ¡:ulrlit knowledge. r\s
lonilal a'¡rai.l,sis ci brrgaini:rg üonte5i5
chapter i I nlalq.es ci*iir il¡ the eotitr:xt of siniple barterirrg. tl-re fact tliat con-
cerieinly it i¡¡ulci lreeÍn í¡remati".ii:e l0 pontiñeate on the shape ti'iat the
tracts have to be rvritten in terms of ovlr¡.t. cen acrually be w*igired or measured
general theor-,¡ will ta-ke r:ther th*i: to obsen'""c that ii rvill l;r-' a ihegry o{' i
c¡"n ¡rtake a very grÉat rieal ni'clifferen¡,:ef - in la¡:i. ell the JifleLende betv"eell
great signilicanle.
a Walrasian r-){.itcome an'd. the Nasir b:rgaining r:cil.liiot.i.
Instead, \\,e tilrü irom rhe sitr,ration in ivhich plai'eis have infcrrnation
'fhls returns us ti:r the clornai¡r of gamr's ui i:i¡raliy we wish tr¡ enlarge on t}-le poiut ¡na.de above iibc-i;t tlte misreprestn-
whicir is secr*t frorn ea¿h other
lation of utiiíty func1.icils. I{L}.,'; dcer hlorn'iati!.ir} ¿ibsui pref'erences *.t'.nl.)
c{}rnpiet{ lnfcrmatioyl in $liich the ¡ules ol the game and the tastc.c anrl
incorpcrated in a i.roriS'ot'contl"ltt-r¡'t t-ir putrlii: kn*wledi¡cl Fresun-ralrl1", iill:
br:iiets cf the players are ccnfton kncv,'ledge. But this retuln to safer
ground
dl'*s n¡.rt eli¡ninaie in{osnaticnal questitllis altogether. Because an itr:m oi story is that prlay'trs iirquire lhis infbrtrratio:r l¡v cbseruing ihe actii:nl cf'
it ric¡es not ntcessarily follow that it il¡
rotii|non knawledge. otheis and nhserving tliaÍ 'rt}¡e¡s are carryiug ou'" siftiiar obcervatio¡ts' Ptssibly'
in1r.i¡¡¡¡¿rtion ig
laltel lreacting, we pl()pose il¡ iircluele ail inf*r- social staiistic¡ máj1,, be av:lilable for ex¡mintllol¡. Fiov,'sver', in nitisl irf tire
ptiblit.kr.au..!É,dg¡, I-i¡1¡1,g¡ tire
ar:tivities being otrserv*cl, either 1iilectl-v or i*dirilctiy, the individuals undei
rl;liftii rvhiii, ;2r bt takell Í-or granted in the draf¡u1-e {ii iontrects' This
ilrclurler ¡r:,:i": r*att*is irs ttre lUrgtiage ín which tfie plai';xs r:l:ninr¡*icafe and observation will not rrgarii tli*nrrel'¡es as fact'less alld *,i1i i:e lveii alvare th¿t
1r,r,11c1:11¡ l¿ct! JrbOul tire wi¡rid but r.''e are c6n¿erned wi¡ir
iníorrrl*ti¡n about
34 And, if any ¡rri:tencc at ¡¿alisnl i$ lil be mainiaa¡¡ed. i).t't'cl!' Freat cost.
tlit'pr'rr.trci'r wirjch ;:ln ire saiij tt l.le pr:blic. in the iast resoft, tl-lis rttf;rns
in1'or-
1.5 ls this ther:ry itcell pirbiic kno,., leCge?
nlati(ln rvh¡cir ¿¿lrl be,¡stalrlisheCi Ll e court of law tJr:ring ail actir¡r fbr hreach 26 Thc la;ge literaturc on i¡centive r:r¡¡np¿tibilit,v and soclal choice is relel¡rnt hete.
oi ¡:t-¡lttr'¡c1. See Dasgupta. -[-ia¡rmonel and l,larkin (1;?v) and Myerson 11984).
ir, : f'or¡nal irf1ri11"!*nt ol'¡iris icsa. the lawyers wcu¡ri i-'e rnocleilr,:ri nt¡t as 27 Tlri¡i is only apparently at varia*cr; wiih what i{arsrnvi (19?Tlencl otl¡e¡r lrave t¡
(.Jt?L_',i .\i::t .11..,::!.,,s of ljl,::: cu¡. hu{ as ntlnilicss auto¡11ilJ ,'x('truiing 3 lay atrout restrirting attenticn,r.ü s{ilufi,.'}ir¡ úüncepts'.vh.icli depend only t:n genuine[y
"l¿atne-theüre1rc va¡i¿lbles'. 5u{rh ais{:rt1,)n} are nlade un the irrrplicii assunption ihai itt¡:
giinrer !s pla'recl iir ísolation - i-e. indeltnrienll¡z of eriiein¿l linrtittlli{r1!s 1-i:ur liarsan.vi
i I T* ¡:i:nil,::t j-.rit e fr:rv. Ch:rtierjee ancl SamueLson (i983). Ctamtor''(t979'
{tr984),
quotes teieirllcne ltumbers as an extraneous variable aithough it is cibvious til¿t sítu¿lions
i::*{:::ri-ni:rg et el il9fll}. h{atthe\r'5 (1983). Mookherjee (1983). \15'"erscn 1982)'
exist fo¡ ruhich knowledge of a certaln l.elephone numtrer ntay be ihe r:rur.:iai i¿ctor.
:li¡r: rr*1s,rn {I? 84 } :-1cb,ri end'tr'ak¡}ri:slii i 1 9 8 3 ).
Nash haryaiaing tkettn': sn introduc{ion Nash bargaífiing theo4,': ün intraduction 2l
l.!reir currenI bchavir.¡ur mü,r' !¡fl¡¡st'l.e the ganies they have ro play in the A Nash equilibrrum is essentialiv a static il{)1ion - i.e. it is s:lti:iai:tcty
future with thrse art.'und tlle¡n. ['hr lact t]rilr tirey are observed v,,iil therelore witliaut refilrernenl30 only tbr ga¡res in which tirne is irrelevant b*cause cach
chaflge thei¡ beliaviour since they will need io strike a balance between player nra¡, tre regarded as rnaking a sltlgle ciroice in ignrlrance of the choic*s
cirirelrt gains arrd potentiai ir-rture lÜsses. Where the futr¡re outrveighs tlie c,f tli*: {:}ther players. As we h¿r,e obsi:rvcil earlier, such static garnes arisc
pre-{dnl, a 'xn¡e' picture of th* piefeLences oi e ur-.pulation may ne!eí *fiierge. natuially in a bargaiiring ciinli"rxt when ctxtltttitinent is possible iluring 1]ie
indred, il an agent systematically reveals'faise' prei'erences, then, fronr an iletgotuitious. However. in stallc bargaining contestS there is an unlbltunale
o¡lerationai polrrt ol view. these ¿re ilis preterences. Eiven the agent hinrself tenelency tbr the set of Nash equrliirria to bc enibarrassingly large. A cowen'
fespeciailv iia politician) mal,ncit be able to disiinguish at aii clearly betrveerr fio¡¡ is rlien rrecessaly to cletermine v,'hich ol these equilibria is t0 tre selected
his'tnre'prelerences and those he has chcsr'n to reveal cver a lon¡r -¡:eiiotl. as th¿ s,llr.rtion (se e chapter 2).
Chapters i0 and i I examine such nl,:de ls in tire context cif'siii;l-ril er:chluge b-or tlynanric garllfs'il {or for gar"nes vo'tiich are forrnr'lliz static but which
ccú;ronlies. The models are v¿ry simple. protiiems r:l cr,ri¡ur.titin,¡r¡. íit"ire and lre to be anaiysecl rviih soure implicit dynamic story in nlrnd). selten (1975)
i*coru¡rlete üifir¡r¡ration being rutirlessly rruncated. É1u1. th.¡iL ltrui:iur"- is has sirorv¡'l lhat refinerttents o1 the lrlasir equilibriurri idea arc necessary'
adequate to demo¡rstrate that, u'hen public knolvleclge ls restricted to Lire Conside;:ation ol these refinemi"rlts iras prov'cd very proi"ltabh: in barfilrining
conimodity b¡lndles actuall-1, barte¡ed (rather 1.har: the vr.:n lieurnann and fheoi1.. The crur;tai papef on thi; topic ls lhat of Rubinstel:r (cliapter 3 of
I!{t-rrgenstern ut.ilities ¿ttaeheo b},tl're ag*nts to t.hese hulrcles), tiren the tiiis v¿¡lum¿). ln sho¡t, 1.he usc l¡i Iefinements of the I{ash eqriiiibriurn idea
hargaining p¡ocess is likely ro zero ir cl¡ a wahasilan ecluilihriunr - even irr q,hirh erpliciti.v take inlo accoullt the d¡rnatrrics of 1h"- :1+qotiation 8anle
the case when there are only'two agenrs anci so neither is negligib,l* cc;mparei.l can resolve altogether the pioblenr of rrlr;ltiple equiliirr"iu l't:r a lairl-v wide
class ol 11,¡io-person bargaining pro'bltmi,. I-'lire* '-lyrianiic *qLiiiii::rium
rvith the total size ol the 'rn¡i¡iret'. Chapter 1 I glosses the se results ir¡' pro','irJ, itic¿:s
ing an ilxicmatic justif ication of a Walrasi¡n e quilibrium wliich ¡rii¡¡ics Nash's rvhich h;¡r,e ireen rvidely used are listed heio1,\,.
axicr¡natic defence ol the Nash bargaining sohltion e-\c('{}í in respe,:t r,i lhe
natüre of the infsrmaticn needrd to compute rhe bar¡lairring olltci)'"e. 1 ¡ieriect e quililiriitn; { Selterr iq75 )' ¡\ loiash equilibriunt
,si,r&gc$le-
11,]-,icli indirces a Nasl'" eq¡iiibrium 1n e','er¡. sufrgill"i.Ie'
) Tremhlitt6'hinzcl'¡:et!'ecí e qtt!littrir,tm r,Scilil¡i 1q?5 i' A limi1 of
4 THL ANIALYSJS cF j\EGOTiATIC¡{ f,Üi\-'fL]STS \asli equlli'nlii oi.perturbed gamls ill v",ilj¡;li t.'tci] ¿g,:lt is supposeci itl
tremirle in rirrkitl¡i decisi,:¡ns so thÍrL each i:v¿jlable ¡cilorr is chcscn witi:
R':cail that we definerl a coiriesi a.r ü game uhich is to be anaiys,-d irr thr to zero"
n I;liiri¡nuin ilro[;]biiit]" e )'0 ¡'l¡ich il then :llor'';ed iÚ recedc -lr¡ spec;if v a
abseiice cf ;rrrl pre-play cornrnunicatli)i-¡ at all. A m!ni¡natr necc.ssarr.ccncition {i8l)'
iiri:i:nl,candidate for the solution of .cucir ¡ giinte rcust :iatisfi'is tlitt ii irc:
J Ser¡uential. ec¡uíiibritmt {K.repr arrrl Wiison I
30 $il-hility consideraiiot.¡s require postulliitig sonre ,1¡ n:rnlics r:r'i'll it 1l¡''!r' lLrIl:
28 Strictly a'gailre of inr:orrplete infc¡nnation'ír ¡r,t :t !ürtj i,.i:ll i:', th,: r¡nse ol vLtn lrnl',. Lrrplicitly irl the viflgs. This is n,:l tc rilr be; got,c j:1.4 il] i]rsttl.
5:¡}, ti1a1 it me"r.
\r'u¡lrann :nd N!,¡¡¡.'¡u,.' ¡1 late ¡cne background dr':nernics: onli tl¡at il is necesslr¡ to tre:¡r:a¡e
thal ulch a
29 .,\lthough il scems hrrsh ':hai lh.¡ itrrr:r¡:nd lJ¿r rs shrulrl l,i: i,i¿nred io¡ so mucl-r pcsrulnte has been made and to he rc¡'Jl'to det'end it
oi which he rv;rs qrrir.e itintce nt. 3 j Bi *'híi:h wi ¡Í:ean a ¡ll;ltt r"'hich is nct staltc'
r1
X,¡as& furr¿'rr*n#ry theory,; an ínftt¡ducti*¡t
their teeth sufltrcieriilv irlts rhe tlynamics Ll u s lt b argain ing t il e ory' tt in ! yt¡ ¿l uc {kt n
." ¿t
l1
tü piovide ¡nuch áLarsráílir€ \\rth the ofproblem
games cf imperfecr in{.orrnatio¡r
of murti¡ri* equiribria. that n players hargain over 1he irivision of a co;ripacr, ür;nvex
-.-supposc cake
T'rembl"ng"hand equilibria a¡e technicalty
iifficutt to cleai wirh and there ''f in utiiity space. Given ¿ i:efererice pr:i't,:i3 c .f anrl positive .tr:rrgaining
{
:j::::T:l,l11,of llrerrrcrarir,:n
when irre results depend on rhe retaiive po.wers' fit,82, .. ., 0*, n:: can clefine an (asvrnmetric) Nash barlau,,ng
r{c({uencres wr¡¡i which egents rnake trembles. sequeniiar
equiiibria present
soiution as the point s e g" ar rvhich
th* ¡rrobleni tl¡¿{ ane has to explain where
the beliefs came from since the (x, -- tr)Fr(;r, _ gr1Ér .. .(,r, _ {,)0,
results can be ven, sensitive to
beiiefs hekl at inf'rmation sets wi:ich
reached in equiribrium."R.ubins¡ein &re nevef is nr¿uiimized subject to the co¡lstraints -x ) and x e s{.
(rgs3) is instructive on this point in so ! This iun has
far as this relates to his chrpier g. been cha¡acterized Lry Roth
{197g) with a set of.thrce axioms: 'of
It rnay not be ¡::ut of o¡'der to speculate that some
at reast .sf ti¡ese r.iffi- Axiotn I : strong indiviclual rationality I
culties are due {c a reruciance to
face the issue of how prayers frnd r}reir
wily Axiom 2: indepenclence of utility calibration;
ro an equilbrium" At ieast tw. routes
can be disringuisheci which.#JJ? Axiom 3: inde pencience of irrelevant alterna rives.
c¡ibecf as educfive anc.*'oful*,e
in Binmore (lgss). The rorrner ref."ers ta
trsditional r*uts described by v.n t'e A function which'raps rrcnr pairs i.rr, g) to poi,ts of
Neun,.rr'and Morgeristern in *,irich the .f and satisiies these
equilibrating process tak*s place inside three :rxioms is necessarilv an asvmnretric Nash bargaining
the pruyrrr, heads. The ratter refers sorutirn frlr sr.,nc
to situatio¡rs in *'hicrr equiribrium is reachel set crf bargaining pow'ers
&. fu,. "",0u. \¡/itir the nclclitic¡n ol a firu¡ih axir¡rn
via an adjustn"lent pr.cess of
trial and er¡or. As we ernphasiaed in poslrlatin..3 svmmetry. the classical lrlash trar.gaiiring
soluii.n is obtaincd
the introducti,rn to the conlpanion
Econo*¿ir: organi:ations *r corrrr, game 6vith Pr -, g? ,:- . . . ,= Én).
l:lr]** theorisis nornra*v def.end
thi"ír anai'sin educriv*ly' and rnucl',""irriJ"'..""(;"il;r-rl?i'-n The Nash brrgai'ini: s'ruiion is p,;ruasive |l irrir voru*re
*_* but exr_.essive
atxempting an evcrlutive ane-trysis
rvithout treating tire backgrc,unJ .,:i.nn*,i* signific:nce shoultl nr¡t bc ailached to ihisj'l a¡: ¡ts 0ori:iranf reappear¿.inrr
arleqr-ratelv. Ltui, l¡ tile c¡_.rntext carl be atiribuir'l l.o tlrr'iri¡1!lli, llil;l¡llilied naturc t-¡f ¡iil. rnudels
cf the Nash program, a¡ edrutive ap¡:ri:ach stu¡lied. ld¡rrc
does noi elwavs t_.arry convicrion in partir:ular fhe i¡lfornr:¡¡;onal assuinpiii_rn tli;¡[ iirc orLlcr.;ln* r;f"ncgotiatir-;ns
because i, or,!¡l oo,"*
rneasure of pro-piai'comrcun.icafio¡r.-u? de1r,i:nds i:¡r'¡/i' *r¡ ,f, an<i f Tiiil rre;rvy ¡il.ci.ir¡¡rtiüÍr
But the spirit",;u,"il";;;uuni,,,
of the i{alh prergra* is " h¡s recciveri iittre attentlon
t'r nlodel sucli ¡xe-pia;' ccmmuni'-atir:n prr:surirablv l-o*causr.: i'l is not lisied as a ii:rmal;:xir¡¡¡l ,
faixrariS,- once this lr*. ü**r, ,i.oru, Br.rt, a:r rvfl i¡r¡¡.rreLi in
one is lefi ivith n*thing erse to iv*rk.¡,'itir sectiorr j. pulilic k-nr¡,¿,iedge is i-rnlikeiy ro i*elutlr piec.ist iníi¡¡¡¡aii,--¡
e;xtepí an *lrciu'ive apprr:ach, r,¡,
Flc¡rvever. recsnt progress in u tilitres.
matireniati;ar hiology sr¡ggrst$ tll:it itLis
rnay
such a i:ad thi*g pro';icied *¡e is sorilril:ing alsr¡ fieeiis to be said ¡.i-.o¡"rt r:riix¡i I rvhiefr hes i:ecn tlie
l]"t ?1 rrüt tempted lo clieat the dynamics
-, subjerl
(see ft,laynerd.Sruirh 19g2. Selren i9g3). 'ri )
ol nru,-,ir ,iehair although selcit¡m iri tile c{rfifext o{ thr:
},J.ash ¡rrogranl . con-
sirii;r a s'l;¡irc bargaining conte$t {sur:}i rs wi,ul<.1 be obtairi*d ,,uith
n,, ,rrn-
strüifiis ,¡n *onmitrnent). Suppose 1ii¡r ¡i:c plavers ciloose straleaics
1.¡*r¡
j seir ,S,,,$?. .. ,5, whici.r are inileSreritleilf .f' ;lnrl {. A funrfr.n ó:,!1 :,
TFIE NASH B"i\,RCAift..tNC SOt.i,irIO¡i ...x 5u 'í. ,T.he pay*fl, r¡e;1ur ii ptrl
+ l-Rr r.let*nnines 'a¡¡reemeilt' pa¡,.r;,iG.
provir.lecl that a:{s)€ .41 . Otherrvise ,tlisagr*ement'
r,:suils pri:iiuciirg ;r p-t\,r)tri.
so f"ar *n tire non-cooperative sitte r'*ctor n'liir.:h wr may take to be { lor simi:licir1, (al:l.rou¡il tiris
L:j::::_:::::i::j]:: of the F,iásli pro{¡aff ¡s nor st¡Ltj:l
L'LL.duls t;irl i¡ 1¡,nere reten: pr\)grCsS
haS been maCle an,l s.here tll, p:un¡r.,a,,, necessar)'for tire argumcnt). Ass¡-r¡ne that this b:rreaining
fbr ir¡medi¿te adva¡ces seern to be . I{owever. cL\rltesr;:lrr,a.r,s ir¡s
sometiring needs to he sald to an rtncquivocal solution in th* se*se that llie solution is a
expiain" the rpparent pervasiveners l",Jasi¡ equiiihríurn
of the Nash bargaíning soite¡.r¡n. {:or ¡:
reasonably cr:mprehensive accoünt ol
cooperative solut¡e.¡i'r.,n.upru taihrecl ln this
to the bargaining problenr, see R.oth 3;1 lltth l'19?:i) irr avcid.inr tile misleaCing ilescriirtio¡ r:i
ctrlirpter, we follor¡.'
1g79i.
f ! ;rs lirt 'slatus q¡jo' point.
-{ec the co¡:r:!ilrling ¡em:ihs trr !¡,,rj¡,¡1 ;. I cr 111.. s:lln,:
feiison. \\c avoid speaking of a nrod*:l rr'iih'i:i-red ihrr*tr'. Bütlt t¿.:rmc inviir:
rnisl*ariin¡
ímplicit assumptions abc!ut commi¡¡tent nossibilities.
32 The fact üat tr.o -. """'::'"..:::*^:'_1'
players know that both 34 lro significa¡ce at ali should be assiglcd to thc asseitiün th;it cn ethír:al nctiir¡ ll.r:¡¡:
t':': or
of rh
them are
ue ,..'
ttsiir-e the same giine fheóry
h,-ór. '-.... ,,^
rL,:rh r¡ir, reqaroedas a ft¡:m of implicit cominunic¿iiolr.- scciatr choice theo¡v has emerged fiom a strategic analysis. lüa¡;tr il':150 did
er¡rticitli.assumed to msrt .ín tl,e ha¡ ** ¡ut ofier,.-piur*r, ,r, I ric.i inilrir:i
f r**;oul; nia¡¡, ]-:is;t*li¿rl as an ethicai iciea and tire¡e seems no p,ocd ¡*¡son rbi *jekir-rA to r*inlilrrci
a,riom -l L.r this direüt;on
':4 Nash bwgrtittirr& ikttlrv: d¡? i¡¡fr(rJ¿df¡dofi Nasfl h*rg*ining tiir:cr1': art introtittttío¡i 15
which Pareto-dúnlirlates anv c¡thrr llasir e,:luiliirria. Suppose iliit the $,oii.íriL,n I-]e ittcnil, G. 19?1:,lf,r¡,{afu¡¡¿.{: lfono;str!-t'f'¿vter Versus Llnio¡¡ Pav't¡' lvll't
s* oi tire hargaining aolltest based on .{'and f yields the agr$eil1ent oul.come Press, Canl t¡ridge, M ;rssar:itLrsctts.
'{s*)} {. trVl-iat happens if # is rrpiaced t_ry I sui:set 1& cuirtaining { and Eilsberg, D. 19?5: 'f i"re theory orrd ¡rractir* of blackmaii. Burgeíníng: Forma!
q:{.s*) as r:r.:iitenrplate,:i in axiom -l'l ühvrously. ,r* remailis a Nash eqili}ibriunr 'I'heories oJ'Neg,otiatlcn. University r:l lilinois Presl, Urbana. Illinois.
rvhich Fareto-donrinates anv other Nash equililrria in ihe nerv gam* and Foldes, L. 19ó;4: A derterminate rnodel r¡i bilateral ntonopoly. Econainica,
therefb e is lire solution of this garne and hence rve irave a jusrific:ltion ol 31, 117-3 r.
u¿iorn 3. in rlynamic bargaining c$n{ests ('l .e. where tlne plays a role}. the F¡ied¡nan, J. 19'77: Atigo¡taly an¿l the 'fheorv of ()arnes. |iorth I{oil¿nci,
A¡nstcrCarn.
s,r¡lumenf abüvc nceds io be adapteii br¡t stiil survives in a modified form.3s
Irudenberg" D., Levine, D. anci 'Tirole , J. i983: Infinite-ho¡1¿on mr¡deis of
F'inaliy. il is necessan¡ to ecnlnlenr on the fact that ncne of the iron-
bargainirg with one-sided inco.nrplete i¡rfcnnation. Forthcoming in [toih,
{rcoperative bargaining rnodels v;hic}i have beeir iitudi*d rrn¡.h::lent the t,jash
A. {ed.) Garne'Theoretic Models rt.l'Rcrgainíng. CUP, Carnb¡idge.
i:a::gainiirg solution exartl..1,. In each cese, the iinplemeniarl cn is approxinicre Cale. D. i1)86: Bargaintng and conlpetilion. i and II. L'cantsmelríca. 54,
{or exaci only in tlie iísrit). ?85-,80{i, 80?-18.
Harsanyi. J. 197'7: Rational lllehat,iour and Bargainür¡ Er;rriirlrrur¡r in f¡:r¡¡nrs
l:i One needs th¡.| une<¡uiv+cal solution-r to subgames irvolvi¡g th* shrunken cake ¿¡id $r¡t:¿al Situations. CLrP, Cambridgr:.
.fravailabii: at time f exist ¿¿nC a¡e simi:ll related ic the soh¡ticn at time 0. l'his is¡l+¡ ilarsan.¡i, .I. and Selten, ll.. 1972: A geürralised N¡¡h s¡luticn fcr twc p;:ison
irue of the Rubinstein model (chapter l\ !¡ecause the $olution depends cn ra'hich playei
ba;gaining garnes v.'.lth incomplete info¡:natir:n" kÍun. -tri., 1S. 80- 10b
gocs fitst except in the li¡nit as tlle leñrlh of the time inie*'als is allorveel io recede to
Hichs, J. 1934: T'he 'I'h'.:or¡' o.f Wtag*'s. S{ai:rliiilan, I"rndon.
zf ro.
tr(reps, i). and Wrlson, R. 198i¡: Sequential eqr:iiiirrium. E*onametrica, 5$,
I {;3-t4.
REFLiRL-iqCES KreÍ:s, I.¡. anri Wilsnn,R. 1981b: F{epr.rt:¡iir:n and imperieci inl'orrnatio:rr
L Econ. Th.,27,253-79.
Aunrann, R.. 1976: A.grecing to dir::lgr,:e. Am. Sta¡..,4. i?-l{t*t). L¿.katos, 1,. 1t)1 5: Froofs antl P-:!'¡títz!irr¡¡.r.',jJie.l.ng¡r r;.r ltslh¡"w*ti''"tii
Binrirore, K" i9'77: Bargaini:rg, ba:rer and competitive etluilibrri.lrn, i and tf. !..i i s c ov e ry, ü r"iP, ii-l;im bridge.
Mirn*o. f-ondon $chooi cl lic*ncn:.ics. ,"04¿ll,ennan. A. i!80: Ageneraiilt' rl-ccoí1ei3Íivr-'1¡¡tont of fi:;painilrg lrr¡rktng
Birrm.]rs, lí. 1983: Bargaining anti coalitions. iCEli"D Jiseussion paper B3¡ I i^ p:iper. linir ersit',i of 'tr^r';'lc'. (.inta¡it-r.
{ ,,¡.lor .: chocl ol }.r'1r¡¡,rlc5. ,10'{athews, S. i!81: Sciling tc ii¡k av*is* b.r5'e ls wirli i,;:,:l'sr;';eb!* f 't¡;tr r.
üinrni.rre. K. 1985: Modelling rational pla-vers. Fortticoming as; lcEf{!-; pape;. J. Et:on. Ti1..'29.
l;r¡,¡l ¡¡¡ Sehoci .lf Fc.r;;omics. I'1a5'ri::rC 5:r:lith, J. i9Bl: lpr¡Ji;¡i¿:¡¡: l¡¡"1 i.!r
Ilinnic-'re. i{. and Dasgu¡rta P. (eci.}, 198Ér: i"r'¿r¡lorlrc ürga,nj:ali.r¡ns *t i'.;¡rc.r. l"--a;n'iiritig*.
* a:i I B lacir',r,ell Ox f¡-¡rd. &dr:rt¡,:ns. .1 . i¡nd Zainii, S. 198.3: Xrcr.n¡¡ii:¡1.i,:)t] '.:'; Il:ii::4111 '' '
it{}trr'1i] |t '1'rj':r':
"
Binr:ror*, K. and I'ie¡rero, M. i 96i : hlatctiing a.nd trargaining in clyn;*,niu an{ 'consis"re¡¡:y' in gi}lnes 'ilrth i:rcon;pltt.e it¡ií¡lil-rü1:r:r'¡ l ir'rf{l ilr¡' i¡:'''t1r¡
;narket:. I{ltlRD cisc::s:;ion pap,:r 8_511 17, I-on{.iou School of Ii,i:onc.rnics. p ap sr, Université C atholiqu r ile L, i iriv¿iii:.
d'hattrr¡e,:. K. :nd Sernueisrn. \\¡_ 1:]E3: Bartr¡ainin.r under inco¡nple te infoí' lr{ocktie:jee. D" i 983' One stage si¡;ii, ii;:ne ':ui i\rrr¡ldii;:ir¡rr '* ri}:' rtt" ':!i , ir L
iirli;rrn;¡ticn. Resea¡cit ¡ra¡*r .-1':)2" lllr;¡*¡cl {:r'iliJr;!* !' '¡'l i¡l {i ':rr¡:r"r"'
!r'
nr:¡tion" üp. Rtst:o,rch, 31. 835-:; i
l:,:,4i.. R i9ó0. j'ltc prcblem oi scciaj üost"J. /,¿:¡+'St..¡;,¡;.. l. {. eiiiornia"
i"lr¿i¡nxi¡n, P. ó9s'+: J'i¡ns and infr:n ation in liargainrrp.. piri] thesjs, $t;infr,rd h{}'erson. R. 1982: Cr:op*raiivr,' gai:i*r ','. il1"' ir:r:rl¡rli;llit ;'rri ¡:r'¡ .lltlin ld.c:r¡r" I