Knowledge Creation with Parallel Teams: Design of Incentives and the Role of Collaboration

Transcription

1 Association fo nfomation Systms AS Elctonic Libay (ASL) AMCS 2009 Pocdings Amicas Confnc on nfomation Systms (AMCS) 2009 Knowldg Cation with Paalll Tams: Dsign of ncntivs and th Rol of Collaboation Shanka Sundasan Th Stat Univsity of Nw Jsy, Zuopng Zhang SUNY Plattsbugh, Follow this and additional woks at: Rcommndd Citation Sundasan, Shanka and Zhang, Zuopng, "Knowldg Cation with Paalll Tams: Dsign of ncntivs and th Rol of Collaboation" (2009) AMCS 2009 Pocdings This matial is bought to you by th Amicas Confnc on nfomation Systms (AMCS) at AS Elctonic Libay (ASL) t has bn accptd fo inclusion in AMCS 2009 Pocdings by an authoizd administato of AS Elctonic Libay (ASL) Fo mo infomation, plas contact libay@aisntog

2 Sundasan, t al Knowldg Cation with Paalll Tams Knowldg Cation with Paalll Tams: Dsign of ncntivs and th Rol of Collaboation Shanka Sundasan Rutgs, Th Stat Univsity of Nw Jsy sundas@cabutgsdu Zuopng (Justin) Zhang Stat Univsity of Nw Yok at Plattsbugh zzhan001@plattsbughdu ABSTRACT Paalll tam statgy has bn widly adoptd by high-tch industis in knowldg cation n this sach, w study th dsign of oganizational incntivs, including a fixd wag paymnt and an additional wad stuctu, fo th paalll tam statgy W consid two typs of paalll tams collaboativ and non-collaboativ paalll tams Poposing and invstigating two typs of oganizational wad policis (individual and agggat) fo both collaboativ and noncollaboativ paalll tams, w dmonstat th viability and chaactistics of ths policis and analyz th tadoff btwn th numb of paalll tams and thi wads W show that collaboation in paalll tams is vital fo obtaining maximal bnfits This sach povids valuabl insights fo fims in mploying paalll tam statgy fo knowldg cation Kywods Collaboation, ncntivs, Knowldg Cation, Paalll Tam NTRODUCTON Tam has bn considd as an ffctiv oganizational stuctu fo all sots of innovativ activitis n a gound-baking study of th pactics of nal Motos in 1943, Pt Duck pointd out th ffctivnss of tam-basd stuctus in oganizations As vidncd by numous succssful cass, tam stuctus hav bn mployd in many nw pojcts Fo xampl, Micosoft launchd its band-nw gaming platfom Xbox to compt with th popula PlayStaion 2 fom Sony, within a shot piod of tim by mploying th tam statgy At oogl, vaious tams a assmbld to wok on diffnt pojcts, such as oogl Documnts, oogl Halth, and oogl Chckout Th a many insightful pspctivs on tam composition Duck (1998) idntifis th kinds of tam as th basball tam, th football tam, and tnnis doubls tam Cohn and Baily (1997) catgoiz tams in oganizations into fou typs: wok tams, paalll tams, pojct tams, and managmnt tams Katznback and Smith (2005) suggst th typs of tams: tams that commnd things, tams that mak o do things, and tams that un things Although thy sm dispaat, ths paps shas simila notions on tams Fo instanc, a paalll tam catgoizd by Cohn and Baily (1997) is th sam as a football tam idntifid by Duck (1998) Du to th incasingly complx natu of tchnologis, not only do fims hav to adopt appopiat tam stuctus, but also assmbl multipl tams to collctivly ngag in a singl R&D pojct Fo instanc, mo than 200 pogamms w involvd at Micosoft to dvlop Windows 95 that contains mo than 11 million lins of cod (Cusumano and Slby, 1997) Thfo, fims a continuously sking cost-ffctiv statgis to manag and coodinat tams on innovativ pojcts Diffnt tam statgis hav bn mployd by industis fo innovation and sach Concunt tam statgy has bn widly adoptd to shotn th dvlopmnt tim of nw poducts Fo instanc, pojct manags at Micosoft usually divid a pojct into pats with spct to its fatus Tams thn wok simultanously on thi pats but thy synchoniz with ach oth and dbug daily (Cusumano, 1997) n contast, with th paalll tam statgy, all tams wok on a sam sach pojct simultanously so as to maximiz th ovall succss at of th pojct Paalll tam statgy is fquntly applid in sach wh singl-tam statgy sults in xtmly high failu ats Fo xampl, Nlson (1961) documnts th adoption of paalll-path statgy in R&D by US Ai foc Simila paalll-path statgy was also cntly usd at National nstitut of Halth(NH) to dvlop malaia vaccin n contast, in th taditional pocss of malaia vaccin dvlopmnt, whn on appoach faild, th ffot did with it as wll (Sigu, 2002) n this pap, w study how th paalll tam statgy can b mployd in knowldg cation n paticula, how should a fim choos th optimal numb of tams to wok on sach simultanously and how to wad th tams? Pocdings of th Fiftnth Amicas Confnc on nfomation Systms, San Fancisco, Califonia August 6 th -9 th

3 Sundasan, t al Knowldg Cation with Paalll Tams Th xists a plthoa of sach on tams in conomics, oganizational scinc, and S litatu Classical pincipalagnt modls xplo th optimal incntiv stuctu und incomplt infomation and th lationship btwn fist bst and scond bst solutions (Holmstom, 1979; ossman and Hat, 1983; Spnc, 1973) Thoy of tams (Mashak and Radn, 1972) quis diffnt incntiv systms (ovs, 1973; Holmstom, 1982; McAf and McMillan, 1991) Knowldg cation in fims follows dynamic pocsss (Nonaka, 1994) with diffnt statgis (Libskind, 1996) Howv, only a fw studis hav analyzd th conomic implications of paalll tam statgis (Aditti and Lvy, 1980; Dutta and Pasad, 1996) Bsids, ths studis do not discuss how to dsign ncssay incntivs to incas th succss at of a sach pojct in conjunction with th optimal numb of tams n addition, th collaboation in tams that affcts th succss at of innovation has nv bn modld bfo Ou sach bidgs this gap by analyzing th citical ol of collaboation and incntivs fo applying th paalll tam statgy Spcifically, w addss th following sach qustions in this pap Fist, how dos th optimal numb of tams and wad policis diff btwn collaboativ and non-collaboativ paalll tams? Multipl paalll tams can b fomd as ith non-collaboativ o collaboativ paalll tams simila to th diffntiation btwn woking goups and tams by Katznback and Smith (2003): non-collaboativ tams is loosly boundd togth fo som common goals, whas collaboativ tams coalsc bcaus of th collaboation among thm n paticula, non-collaboativ tams wok indpndntly without laning fom o shaing with oth tams, whas collaboativ tams wok closly togth so as to ffctivly incas th succss at of th sach pojct W study whth and how th psnc of collaboation in paalll tams hlps a fim to dsign btt incntiv contacts to achiv maximal bnfits Scond, how should a fim dsign incntiv contacts fo paalll sach tams to induc thm to xt thi bst ffots? Th incntiv contacts fo a tam in this sach consist of a fixd wag paymnt and an additional wad policy All th tams gt th fixd paymnt no matt whth thy will succd in sach and thy will b wadd additionally if thi sach pojct succds Two typs of wad policis (individual and agggat) a poposd in this pap and analyzd as to how ths wads should b dsignd to induc th bst ffots fom tams so that th fim may achiv maximal pofit Finally, how many paalll tams should a fim mploy fo knowldg cation? Pio sach studid th optimal numb of paalll tams, but without th incntiv issus This sach maks th significant xtnsion by invstigating th optimal numb of paalll tams whil taking into account th incntivs to motivat th bst ffots Th pap pocds as follows 2 outlins ou modl 3 psnts th dtaild analysis and discussion 4 povids managial insights and concluds th pap MODEL n this sction, w psnt a modl in which a fim dsigns optimal incntivs fo paalll tam statgy Bginning with th individual tam s dcision poblm und th individual tam wad policy, w thn show th oganizational dcision poblm Finally, w div a simplifid modl fo analysis W consid a fim that wishs to mploy multipl paalll tams to ngag in a sach pojct fo knowldg cation Each tam, acting indpndntly, xts a ctain lvl of ffot, gnating a succss at fo th sach pojct n ou modl, th tam is considd a unit of analysis Th spcifics of how th collctiv ffot is distibutd within th tam and how th tam is composd of a discussd in a diffnt sach pap n addition, w assum all th tams a homognous with spct to thi abilitis in sach Th fim chooss M numb of paalll tams and offs a paymnt stuctu that consists of two pats: a fixd wag paymnt w and an additional wad Th paymnt fo individual tam wad policy can b modld as w unsuccssful tams, individual policy= w+ succssful tams Und th individual policy, a tam i s xpctd nt payoff is π = w c( ) + ρ (, M ), (1) i i i Pocdings of th Fiftnth Amicas Confnc on nfomation Systms, San Fancisco, Califonia August 6 th -9 th

4 Sundasan, t al Knowldg Cation with Paalll Tams is its ffot xtd, c( ) is th cost of ffot, and ρ (, M ) is th individual tam s pobability of succss at wh i i which will b discussd in dtails lat on Accodingly, th fim s nt payoff is π = ρ ( E, M ) B M w ρ (, M ), i M i (2) i in which E is a vcto containing all paalll tams ffots, B is th bnfit of th sach pojct, and ρ ( E, M ) is th goup succss at of all th tams B ρ ( ) ρ ( ) U M R w Paamts bnfit of sach individual tam succss at goup succss at of all tams svation utility Dcision Vaiabls tam ffot numb of tams individual tam wad agggat tam wad fixd wag paymnt Tabl 1 Summay of Notation Fomally, th fim s poblm [ P ] can b dfind as subjct to max π w, M, E Equilibium st E, π U, p w 0,, (3) wh th fist constaint is th incntiv-compatibility constaint (C) fo tams in which E is th quilibium st of ffot among M tams, th scond constaint is th individual-ationality constaint (R) in which U is th svation utility, and th last on is to nsu a positiv wag paymnt (s Tabl 1 fo th complt list of notations) Whn th agggat wad policy is applid, all th tams will qually sha an agggat wad R whn th sach succds; th individual tam s payoff and fim s pofit can b fomulatd accodingly W conduct ou analysis in th nxt sction fo diffnt scnaios: (1) fo non-collaboativ and collaboativ paalll tams; (2) fo individual o agggat tam wad policy Fo a tam in a non-collaboativ stup, individual tam succss at dos not colat with th total numb of tams Hnc, th individual tam succss at can b dfind as ρ (, M ) = ρ( ), which is indpndnt of th total numb of tams M Thn th goup tam succss at is just ρ (, M ) = 1 (1 ρ (, M )) M = 1 (1 ρ( )) M t is assumd that 2 ρ ( )(1 ρ( )) + [ ρ ( )] 0 Pocdings of th Fiftnth Amicas Confnc on nfomation Systms, San Fancisco, Califonia August 6 th -9 th

5 Sundasan, t al Knowldg Cation with Paalll Tams Fo a tam in a collaboativ stup, individual tam succss at dpnds on th total numb of tams Du to th collaboation among all th paalll tams, individual tam succss at may incas whn mo tams ngag in sach n this gad, w dfin th individual tam succss at as ( ) ( ) 1 (1 ( )) q M M ρ, M = ρ, thn th goup tam succss at M q( M ) will b ρ (, M ) = 1 (1 ρ (, M )) = 1 (1 ρ( )), wh q( M ) masus th dg of collaboation among M tams Whn q( M ) = M, th is no collaboation among tams, which is th sam as th abov cas fo a goup of tams Whn q( M ) > M, th is collaboation among tams, which hlps to impov individual tam succss at ρ (, M ) Whn q( M ) < M,th total numb of tams will hav opposit and ngativ ffct on individual tam succss at, which vntually ducs individual tam succss at ρ (, M ) n gnal, w assum that q( M ) is a concav function, fist incasing thn dcasing in M with th poptis that q (0) = 0 and q (1) = 1 W nxt intoduc two lmmas, which gatly simplifis th fim s poblm Lmma 1 Th always xists a uniqu symmtic quilibium of tams ffot lvls whn th individual wad policy is applid Poof Plas s Appndix A Lmma 2 Th individual-ationality constaints a always binding fo ach tam Poof Plas s Appndix B Following upon Lmma 1, w simplify th notation i as fo all th individual tam ffot lvls und th individual wad policy and th incntiv-compatibility constaint is thus ducd into agmax{ w c( % ) + ρ ( %, M ) } Lmma 2 lts us futh duc th fim s poblm with individual wad policy into subjct to (, M ) % max π = M[ c( ) + U ] + ρ (, M ) B c( ) ρ (, M ) + U 0 { % c ( % ) = ρ ( %, M ) }, % ANALYSS AND DSCUSSON This sction dtails ou analysis on paalll tams Fist, w discuss th optimal solution fo non-collaboativ tams with individual wad policy Scond, w xplo vaious conditions focusing on th tam lasticity of collaboation fo collaboativ tams Finally, th impacts of agggat tam wad policy a invstigatd on th quilibia of tams ffot lvls Non-collaboativ tams Fo non-collaboativ tams, w show in th nxt poposition that th fim can only achiv th scond-bst solution by not offing any fixd wag paymnts to tams bcaus of th lack of collaboation in tams Poposition 1 Th fim cannot achiv th fist-bst solution fo non-collaboativ tams and th scond-bst wad and optimal ffot lvl can b dtmind fom Pocdings of th Fiftnth Amicas Confnc on nfomation Systms, San Fancisco, Califonia August 6 th -9 th

6 Sundasan, t al Knowldg Cation with Paalll Tams Poof Plas s Appndix C c ( ) = ρ ( ) ρ( ) U = c ( ) c( ) ρ ( ) Th ason that th fist-bst solution cannot b achivd in non-collaboativ tams lis in th contadiction that th fim cannot obtain a positiv pofit whil maintaining a positiv wag paymnt Figu 1 xplains this intuition by showing individual tam s iso-utility wag contous and th fim s indiffnc cuv whn th is no collaboation in paalll tams Figu 1 Tam s iso-utility wag contous and fim s indiffnc cuv: no collaboation Sinc individual tam s ffot lvl is solly dtmind by c ( ) / ρ ( ) =, which is not latd to th numb of tams M Thn, d/ dm = 0, which implis that iso-utility wag contous a hoizontal lins Figu 1 shows ths lins fo th svation utility U Th fim s substitution at is which is zo at point ( M ) π d M ρm B w ρ( ) + ( ρ B Mρ ( ) ) M = =, π dm ( ρ B Mρ ( ) ) Mρ( ), that satisfis th following two conditions ρ B= w+ ρ( ) = c( ) + U, M ρ B= Mρ ( ) = Mc ( ) Ths two conditions suggst that th optimal ffot lvl is chaactizd by c ( ) c( ) + U = ρ ( ) (1 ρ ( )) ln(1 ρ ( )) Whn th wag paymnt is zo, individual tam s ffot lvl can b obtaind fom c ( 0 ) c( 0 ) + U = ( ) ( ) ρ ρ 0 0 Sinc (1 ρ ( )) ln(1 ρ ( )) < ρ( ), > 0, it follows that > 0, o > 0, which implis that th fim s indiffnc cuv fo maximal pofit can nv intsct with individual tam s iso-utility wag contous fo utility U Thfo, th fim can only achiv th scond-bst solution by lowing to 0, offing zo wag paymnt to individual tams M mains unchangd in both fist-bst and scond-bst cass n summay, in a goup of tams, th fim Pocdings of th Fiftnth Amicas Confnc on nfomation Systms, San Fancisco, Califonia August 6 th -9 th

7 Sundasan, t al Knowldg Cation with Paalll Tams cannot achiv th fist-bst ffot lvls fom tams Th scond-bst ffot lvls can b obtaind by paying tams a popotion of th sach bnfit und ctain conditions Collaboativ Tams Having illustatd th fim s scond-bst solution to a goup of tams, w nxt tun ou attntion to th incntiv dsign and th optimal numb fo collaboativ tams ndividual wad policy n this sction, w dmonstat that fo collaboativ tams, th fim may achiv a fist-bst solution und som conditions (s Appndix D fo th lagangian function of th fim s pofit) W fist dfin and invstigat th complmntaity btwn th ffot lvl and th total numb M of tams, thn popos th concpt of tam lasticity of collaboation, and finally chaactiz th conditions to obtain th fist-bst and scond-bst solutions Th complmntaity is dfind as th lation btwn th ffot lvl and th numb of tams with spct to thi contibution to th fim s total xpctd pofit f 2 π/ M = 0, th xists no complmntaity btwn and M, 2 which mans that th ffot lvl is indpndnt of th total numb of tams; if π/ M > 0, th xists positiv complmntaity btwn and M, implying that th ffot lvl incass whn th a mo tams; and if 2 π/ M = 0, th xists ngativ complmntaity btwn and M so that th ffot lvl dcass whn mo tams wok in paalll Th nxt lmma shows th sufficint condition fo th Hssian matix of th fim s pofit to b ngativ dfinit, lading to th futh discussion on th complmntaity btwn and M Lmma 3 Th sufficint condition fo th Hssian matix of th fim s pofit to b ngativ dfinit with collaboativ tams is that th following inquality holds at th stationay point (, M ) Poof Plas s Appndix E q( M ) q ( M ) q( M ) ln(1 ρ( )) q ( M )[1 + q( M ) ln(1 ρ( ))] (4) M Basd on th abov lmma, w fist consid th cas whn 1 + q( M ) ln(1 ρ( )) 0 As povd in Appndix E, th always xists ngativ complmntaity fo this cas btwn and M and th abov inquality (4) can b simplifid as q( M ) 1 q ( M ) M Scondly, whn 1 + q( M ) ln(1 ρ( )) > 0, if q( M ) [1+ 2 q( M ) ln(1 ρ( ))] < [1 + q( M ) ln(1 ρ( ))], q ( M ) M th xists positiv complmntaity btwn and M, and if q( M ) [1 + q( M ) ln(1 ρ( ))] < 1, q ( M ) M th xists ngativ complmntaity btwn and M Finally, whn th is no complmntaity btwn th ffot lvl and tam siz M, th Hssian matix is always ngativ dfinit n addition, th following quation always holds dm dq( M ) = [1 + q ( M ) ln (1 ρ ( ))] M q( M ) (5) (6) Pocdings of th Fiftnth Amicas Confnc on nfomation Systms, San Fancisco, Califonia August 6 th -9 th

8 Sundasan, t al Knowldg Cation with Paalll Tams Sinc q (1) = 1, th lationship btwn th optimal ffot lvl and th numb of tams M can b solvd as W dfin th tm ( ) 1 M = q( M )(1 ρ( )) q M (7) Mq ( M ) ψ = q( M ) as th tam lasticity of collaboation which masus how th dg of collaboation changs whn th is on mo tam in th collaboativ stuctu Notic that this tam lasticity can b ngativ if th collaboation stngth stats to dcas whn mo tams a in th goup Th conditions fo th stationay point (, M ) to achiv ith positiv, zo, o ngativ complmntaity a summaizd in Tabl 2 Tabl 2 Complmntaity conditions gading th tam lasticity of collaboation Having discussd th complmntaity btwn and M with spct to tam lasticity of collaboation ψ, w nxt dmonstat th conditions fo th fim to achiv a fist-bst solution Poposition 2 Und individual wad policy, th ncssay and sufficint condition fo both th fim s optimal pofit and th fixd wag paymnt to b positiv is and th tams optimal ffot lvl can b chaactizd as which can b inducd by offing th optimal wad as Poof Plas s Appndix F Mρ q ( M ) M ρ, (1 ρ ) ln(1 ρ ) q( M ) (1 ρ ) ln(1 ρ ) c ( ) q( M ) (1 ( )) q( M ) 1 = ρ B, ρ ( ) M q( M ) c ( ) (1 ρ( )) = = B ( ) ( ) (1 ( )) q M ρ, M M ρ Poposition 2 dmonstats th conditions fo th fim to obtain th fist-bst solution, which nabls th fim to off a positiv fixd wag paymnt and still maintain a positiv optimal pofit Th condition ssntially implis that und th individual wad policy, th ncssay condition fo both a fim s optimal pofit and th fixd wag paymnt to b positiv is th xistnc of ngativ complmntaity btwn and M n oth wods, whn th fim achivs th fist-bst solution, th should always xist th ngativ complmntaity btwn th goup siz and ffot lvl, i, whn mo tams join in th goup, ach tam xts lss ffot Whn th conditions of obtaining th fist-bst solution cannot b satisfid, th fim may still achiv th scond-bst solution und ctain condition, which is psntd in th nxt poposition Poposition 3 Und th individual wad policy, th fim can only achiv th scond-bst solution by paying a zo wag paymnt if Pocdings of th Fiftnth Amicas Confnc on nfomation Systms, San Fancisco, Califonia August 6 th -9 th

9 Sundasan, t al Knowldg Cation with Paalll Tams Mq ( M ) Mρ <, q( M ) (1 ρ ) ln(1 ρ ) and th ffot lvl and wad in this cas can b dtmind fom Poof Plas s Appndix c ( ) =, ρ (, M ) c( ) = ρ (, M ) U Figu 2 dmonstats th iso-utility wag contous fo tams btwn wad and th numb of tams n paticula, fo a fixd wag paymnt w to achiv ctain utility, th can b vaious combinations btwn and M, that is, w i d M ρm = = < 0, w i dm ρ and 2 2 d / dm > 0 Hnc, th substitution at btwn and M fo a fixd wag monotonically dcas Figu 2 Tam s iso-utility indiffnc cuvs fo a fixd wag: fist bst Th fim s maginal indiffnc at btwn and M fo a fixd pofit is π d M ρm B w ρ MρM + ( ρ B Mρ ) M = =, π dm ( ρ B Mρ ) Mρ which will qual individual s substitution at at th point ( M ) l, that satisfis conditions ρ B= w+ ρ = c( ) + U, M ρ B= Mρ = Mc ( ) Ths two conditions a ssntially th fist-od conditions of th fim s pofit with spct to and M f th optimal point ( M ) U, as shown in Figu 2, thn th, is within th gion fo tam s iso-utility contous with utility fist bst solution can b achivd Howv, if th optimal point ( M ), is out of th gion fo tam s iso-utility wag contous with utility U as shown in Figu 3, th fim has to mov its indiffnc cuv downwad such that it is tangnt with tam s iso-utility wag contou at w= 0 n this cas, th fim has to bind tam s positiv wag constaint and achiv a positiv scond-bst pofit Pocdings of th Fiftnth Amicas Confnc on nfomation Systms, San Fancisco, Califonia August 6 th -9 th

10 Sundasan, t al Knowldg Cation with Paalll Tams Figu 3 Tam s iso-utility wag contous and fim s indiffnc cuv: scond bst Agggat wad policy nstad of offing wads to succssful individual tams, th fim can qually wad all th tams whn th sach succds W fist dmonstat a simila symmtic quilibium as that und individual wad policy Und th agggat wad policy, asymmtic quilibium may xist wh tams xt ffots at diffnt lvls Howv, w finally dmonstat that only th symmtic quilibium xist as long as all th tams in sub-tams collaboat as on tam f th fim mploys agggat wad policy fo its paalll collaboativ tams, simila symmtic quilibium may xist as that und th individual wad policy Th nxt poposition shows that no such conditions xist fo th fim to both off a positiv wag paymnt and achiv a positiv pofit Poposition 4 Und th symmtic quilibium of agggat wad policy, th ncssay and sufficint condition fo w 0 (o, fo th fim to b abl to off a positiv fixd wag paymnt) is q ( M ) M Mρ, q( M ) (1 ρ ) ln(1 ρ ) and th ncssay and sufficint condition fo th fim s optimal pofit to b positiv is q ( M ) M ρ, q( M ) (1 ρ ) ln(1 ρ ) and th ffot lvl and agggat wad in this cas can b dtmind fom Poof Plas s Appndix H R c ( ) =, M ρ (, M ) R c( ) = ρ (, M ) U M Poposition 4 illustats that it is impossibl fo th tam lasticity of collaboation to satisfy both conditions as thos fo individual tam wad policy Hnc, th fim cannot achiv a positiv pofit and off a positiv wag paymnt at th sam tim und th symmtic quilibium of agggat wad policy Thfo, to achiv a positiv pofit und th symmtic quilibium of agggat wad policy, th fim should not off a fixd wag paymnt to tams Pocdings of th Fiftnth Amicas Confnc on nfomation Systms, San Fancisco, Califonia August 6 th -9 th

11 Sundasan, t al Knowldg Cation with Paalll Tams W nxt invstigat th possibl asymmtic quilibium whn th xist two o mo sub-tams among paalll tams W show that th optimal solution quis th fim to induc tams to xt ffots at th sam lvl no matt how many sub-tams may xist Poposition 5 Th asymmtic quilibium dos not xist whn th fim offs th agggat wad and all paalll tams collaboat as on tam Poof Plas s Appndix Poposition 5 implis that all th tams xt sam ffot lvls as long as thy njoy th collaboation among all th paticipants Th solution in this cas is th sam as that und th symmtic quilibium f th hadly xists any collaboation among th sub-tams, thn th goup succss at will b diffnt and th may xist asymmtic quilibium among tams Figu 4 Vaious conditions with spct to tam lasticity of collaboation ψ Figu 4 gaphically summaizs ou findings about two typs of wad policis with gad to th tam lasticity of collaboation Accoding to ou conditions in pvious popositions, diffnt aas a idntifid with spct to th tam lasticity of collaboation n addition, w add in collaboation and complmntaity to th chat so that thi lationship to th fim s dcision can b asily sn Sinc th lacks collaboation in non-collaboativ tams, th tam lasticity of collaboation is 1 and th fim can achiv a positiv pofit, but cannot off a positiv wag paymnt to tams Whn th xists no complmntaity btwn tam ffot lvls and th numb of tams, it is possibl to off a positiv wag paymnt to tams und both typs of wad policis; howv, th fim will not b abl to achiv a positiv pofit und such situation Th shadd aa psnts th ang of tam lasticity of collaboation that nabls th fim to achiv th fist-bst solution, both th fim s optimal solution and th fixd wag paymnt a positiv Figu 4 also psnts th compaison btwn individual and agggat wad policy: th fist-bst solution may b obtaind whn th tam lasticity of collaboation is within a ctain ang und individual wad policy, but only scond-bst solution can b achivd und agggat wad policy MANAERAL MPLCATONS AND CONCLUSON Th incasingly comptitiv makt has focd companis to sk mo cost-ffctiv ways to ngag in knowldg cation Th cnt tnd in outsoucing knowldg claly indicats that companis a constantly saching fo th bst businss statgy to not only sav th costs but also impov th quality of knowldg discovis To div impotant managial insights about how to ffctivly mploy paalll tam statgis, w psntd a modl of paalll tams and incntivs in which a fim mploys multipl tams and dsigns incntivs to motivat ths tams to xt thi bst ffots Ou analysis povids valuabl guidanc fo manags in dploying paalll tams as discussd blow Fist, motivating tams to ffctivly ngag in knowldg cation is ssntial fo fims to impov thi poductivity and ovall pfomanc W show how appopiat incntivs can b dsignd Appopiat incntivs (fo instanc, wag paymnts) can b dsignd to induc woks bst ffots in knowldg innovation, nhancing th succss at of knowldg discovy Scond, collaboation is indispnsabl within paalll tams fo knowldg cation to achiving maximal bnfits n Pocdings of th Fiftnth Amicas Confnc on nfomation Systms, San Fancisco, Califonia August 6 th -9 th

12 Sundasan, t al Knowldg Cation with Paalll Tams non-collaboation tams, th fim should not off any fixd paymnt to a goup of tams, but only off th wad pat that shas th knowldg cation bnfit, sinc only th scond-bst solution is attainabl Thid, succssful innovation tams can b wadd ith individually o collctivly Although it is possibl to achiv th fist-bst ffot lvls with th individual tam wad policy, th incntiv to motivat collaboation may not b so stong bcaus only succssful tams gt th wad n contast, und th agggat wad policy, tams will sha th total wad as long as any tam succds, so thy may b inducd to voluntaily collaboat with oth tams This study shds light on how incntivs and collaboation among tams affct oganizational dcisions on knowldg cation W plan to study th unctainty of innovation bnfit with potntial knowldg discovy and invstigat th impacts of infomation tchnology in mo dtail n conclusion, ou pap povids valuabl insights fo manags to choos th bst numb of paalll tams fo knowldg discovy and also dtmin appopiat lvl of wads to achiv optimal oganizational pofits REFERENCES 1 Aditt, F D, and H Lvy (1980) A Modl of th Paalll Tam Statgy in Poduct Dvlopmnt, Th Amican Economic Rviw, 70(5), Cohn, S, and D E Baily (1997) What maks tams wok: goup ffctivnss sach fom th shop floo to th xcutiv suit, Jounal of Managmnt 3 Cusumano, M A (1997) How Micosoft Maks Lag Tams Wok Lik Small Tams, Sloan Managmnt Rviw, 39(1), 9 4 Cusumano, M A, and R W Slby (1997) How Micosoft builds softwa, Communications of th ACM, 40(6), Duck, P (1998) Managing in a Tim of at Chang Plum 6 Dutta, J, and K Pasad (1996) Laning by Obsvation within th Fim, Jounal of Economic Dynamics and Contol, 20, ossman, J S, and O D Hat (1983) An Analysis of th Pincipal-agnt Poblm, Economtica, 51(1), ovs, T (1973) ncntivs in Tams, Economtica, 41(4), Holmstom, B (1979) Moal Hazad and Obsvability, Th Bll Jounal of Economics, 10(1), (1982) Moal Hazad in Tams, Th Bll Jounal of Economics, 13(2), Katznback, J R, and D K Smith (2003) Th Wisdom of Tams: Cating th Highpfomanc Oganization HapCollins (2005) Th Disciplin of Tams, Havad Businss Rviw, Th High Pfomanc Oganization, pp Libskind, J P (1996) Knowldg, Statgy, and th Thoy of th Fim, Statgic Managmnt Jounal, Spcial ssu: Knowldg and th Fim, 17, Machak, J, and R Radn (1972) Economic Thoy of Tam Yal Univsity Pss, Nw Havn, CT 15 McAf, R P, and J McMillan (1991) Optimal Contacts fo Tams, ntnational Economics Rviw, 32(3), Nlson, R R (1961) Unctainty, Laning, and th Economics of Paalll Rsach and Dvlopmnt Effots, Th Rviw of Economics and Statistics, 43(4), Nonaka, (1994) A dynamic thoy of oganizational knowldg cation, Oganization scinc, 5(1), Sigu, E (2002) ntviw with N Rgina Rabinovich: Conductd by Eica Sigu, ntnational Halth, 3(2) 19 Spnc, M (1973) Job Makt Signaling, Th Quatly Jounal of Economics, 87(3), APPENDX Appndix is omittd du to lack of spac and is availabl at Pocdings of th Fiftnth Amicas Confnc on nfomation Systms, San Fancisco, Califonia August 6 th -9 th