Professional Documents
Culture Documents
An Introduction To The Economics of Information Incentives and Contracts 2nd Edition PDF
An Introduction To The Economics of Information Incentives and Contracts 2nd Edition PDF
spettothigame,tisnatraltasime that ech get ttempts to marie
‘sown utiity, so the agents are plavinganon cooperative game or theta
ses that they choose shoul forma Nash equliriun. However this does not
rnean that they will never co-operate, Even ina non-co- operative game if
‘thee exist mechanisms that allow the agents to arrives co-operative aie
‘ments ad to take advantage of them these mechanisms may wel be used A
Second posbty st anayae what happens when the agents Ue co-operate
behaviour anyway. All these considerations influence the third age of the
fame the incentives stage and so by backward induction they alo inence
‘he optimal eotrac.
“The participation tage silo that ofthe simple models asthe con
tract design stage, We sould away bear in ind thatthe contrat intended.
fora parbeular agen may incall the verifiable variables ofthe ene rela
tionship those dre eterringto the agent at hand as wel asthose refering
tothe other agents.
In the literature on optimal contacts with several agents thee exis ifr
‘ent groups of models according to the aspect analysed. The personalized
information models (whore hasic references are HolmsteGm (1979) and
‘Moore (1984) ) anal contracts designed for ach agent using not ony
the ersonalze information butalsothe information on the eu obtained
bythe other agents The principal wil be interested in paying each agent ac-
cording to his owen prodsction and that ofthe other agents if thee othe ce
sults can inform onthe actions ofthe agent at hand the rests ofthe other
agents do not add information, or in other words, if an agents real ina
'sufcienstisicforhs eos then he willbe pid acovding to his own elt.
only
Tet us consider simple example to ilustrate the above mesg. A depart
ment store contratste salesmen, one forthe book department. andthe other
forthe umbrella department. Sle depend on the effort ofthe salesmen and
fn arandom variable, but the random variable that let each departaent is
Aiterent nti ase ts optimal foreach agent tobe pad only aeconding 0
the sales ofhisown department the bok saesman willbe paid scoring 0
the number of books that are sold the umbrella salesman according 0 the
number of umbels The reason ithat the effort ofthe book alesinan does
not affect the sales of umbrellas and these ates do not evea anything on he
book salesman’ efor since the events that mk the sales of umbels high
are diferent fom those tht aft the book markt. Consequentyincluding
‘The Moral Hazard Problem
the sles of umbrellas tone ofthe variables that determines the bookseller’
‘wage can only make the salesman sfler additional risk without motivating
bi farther Now imagine what would happen ifboth salesmen were working
inthe same department, for example, bth sll bres In this ase we ca
‘fly ssumetat the randomnes that affects sales the same for bothagents.
Hence fthe efforts were seit islkly thatthe ress would bealso sim
far butif th eflorts were deen it sliely that he ress would be df
‘ent For example, the fact that it fans is common tothe results of both
Silesmen Since the outcome ofthe random weather variable the wae for
both ine sleaman sllsalotand the other does sel may umbels
Aicultfor the one who has the worst rl to argue that it was dae w bad
Tuck Instead signal that he exerted lester. This ends sto situation
{in which the optimal contac foreach agent shoul take ino account the
sulteofbathof them. The deparment store shoul psy more to aslesman the
leas workmates have sold: However, ii posuibl or the contacto include
{his information inthe oppose dietion. For example it tems reasonable
that we should place a higher value on the sas of an umbrella salesman the
tester the sales of things hate been (high sales of this item signal
thatthe weather as been hot, which san adverse condition forthe umbrella
salesman.
In gencral optimal contracts can become very complicated Sometime,
however, we observa very particular type of contract Think of businesses
that paya bonds tothe salesman who sls the most (the employee ofthe
‘month) This type of pay-off, fr which the only important fctoristhe order
in which the agents place themsles, is known a tournament contact.
Under thistype of contract an agent payoff depends oaly onthe ranking of
Iisresulin the ordering of the reuls feather agents. Sports competitions
are the dearest example of tis typeof contract. Howeve,ourmaments ae
‘only ficient in very particular cases since forthe optimal contact to depend
tonly onthe order ofthe results tisnecesary that all the revant information
besummarized inthis ranking
azar and Rosen (1981), Cen and Stokey (1983, and Malcomson (186)
nays the advantages and disadvantages of the mos simple payment mech
‘snismsthose that depend only onthe individual rev andthose that depend
‘onlyon hens ordering the results Aswehavesen, inspite of the fact that_
‘hse contracts re optialoaly under rather estictive conditions, they ae
‘the most used in practice which juss studying them. A asicconcuson i
that contacts bated only on individual results are batter when the common.
poise (the randomness that simultaneously afc all sul) is sal while
‘Tournament are beter when the common ait is important. Intredcing
lack of commitment o non-stationary environment can also provide c=‘The Moral Hazard Problem
‘umsances where tourmaments are optimal: Malcomson (1984) considersthe
‘ase in which the performances observable by the principal but net vera
tothe courts Now the principal hasan incentive to renege onthe promised
‘bonus. However, she may be able o commit toa tournament where the rie
has tobe paid anyway. Nlebuf and tight (1983) argue that a advantage of
tournaments is that they provide Tele’ incentives the sense that they
‘hangeactomaticaly wth hangesin the environment wile the optimal con
cast be ecaclsted each ine,
“The literature has also considered what are known as join production
‘model in which the final reat depends on teamwork This ic the case of «
_r0up of agents whose individual efforts combine prod a single good,
Examples ofthis typeof situation are cesearch teams that develop «given
produit a group of frm that cary outa common projec, or football eam
Inyingto win championship.
‘When the ial prediction isthe ony weiale variable, ant depends on
the vector of (non-obsertale) efforts rom the perspective ofthe ea of an
‘optimal contract we cannot obtain conclusions that ae very diferent from
those that correspond to models with only one agent The fundamental difer-
‘ence is that group reationsiprare closer tothe radional problem of public
s00ds,andso the problem that arises isoneof fre ridingsinethe effect of any
‘edactionin flor exerted isshared between althe agents, Wend each agent
to fel responsible forall the product in order to provide the right incentives
foci. That why lchian 2nd Dems (1972) and Holmstrom (1982) so
st that situations joint production (ia which tisimpossibetoidentifyhe
dividual contributions cannot beeficient when all the income is distib>
‘ted amongst the agents In other words the budget consaitalwaysbinds,
itisaot possible to achieve efcieny. This means thai certin cases anber
agent shoul be contacted to contr the prodcive agents Ontop ofthis he
{external participant (the principal) can bein charge of making sre that the
‘agreements that do nt balance he Budget (ie that do ot disteute the ou
‘ome in any coningeney among the agents are fulfled The pressure of en
twa and the ole of residual claimant can provoke efcieny gins and this
jsty the presence ofa principal
‘Until now we have always arsed thatthe agent ply non co-operative
theycan co-ordinate, then coalitions form, which can be harfulfor there
inship.” Consider, fo example. a principal who contracts wo agents one
isin charge of the productive aks (we cal him the agent), and the other sin
charge of supervision and contol (the foremas)- Ia this ae, iis not unreal
ca eh esd lpm Se Sol oe
0
I
|
|
|
“The Moral Hazard Problem
‘ot inequent forthe agent to ay this foreman if something goes wrong,
nd you do'leton that was yf, then 1 pay you foto forthe ore
‘man to goto sete agent ant tll i stn, if something goes wrong, if
‘you ke wecan come to an arrangement forme o protect you I this type of
oltion is formed then the formats objective so protect the agent when
the rele are adverse, covering up the evidence and announcing othe ria
‘pal thatthe contol didnot give any condusv sul Therefore oni
{ormation thatthe principal would haves that which favours the agent
"The consequence of this that, when the Principal designs the pay-ofF
mechanism, she will ancpatethat her workers may haveinentives to collde
nd wil ry not oleae ny mary fr coalitions tobe profitable other par-
‘pans, This brings ust an important thought forthe theory of orgie
‘ons. coalitions ate possible the principal cannot kt the agent's punishment
{or prize) depend totally on the pervs. I the objective isto reduce the
‘Stablshment of agreements between the gents and thei supervisors she wll
‘have to apply rules that ae impersonal, uniform (that eliminate the con-
Ssequencesof personal and repeated contrat) and donot seal the informa
tion the source ofthe distortions) as is done in bureaucratic rues. The
_ssge that when the indvidal in a organization hae enough leeway,
‘We ca find an argument in favour ofthe technical uperonty of buresucracy
‘over oer forms of organization. This superiority is based on its precision,
ontinity andthe uniformity of is norms,whichsaeson costs and tions,
nd voidethe formation of income-seking coalitions. However inthe same
‘way the possiblity hat coalitions may exis unambiguovsy implies that there
‘san fen loss inthe operation ofthe organization,
"Arcatively widespread esl thatthe more leeway the agents have the
more complex the situation under consideration then the es that re at
posed tend to besinpler Theapplication of bureaucratic rules however not
{he only solution. A second possible solution tothe problem of eceney
losses due to coalition formation betwen supervisors and agentssto system
tially ue the services of external auditors o supervisors (ee Kofman and
‘aware, 1993). This sytem is costly since the external supervisor must be
pai but tallows the establishment of control and punishment mechan-
{sm that issuades the participants in the hierarchy from adopting coliosive
behaviour
‘The conchsion ofthe models that we have just considered is rather pes
simistie However ts ety to think of example in which co-ordination can
Jimprvetheinputofa group of agents, andin wich the advantagesobreau
«racyaeoniy availble when tisimporantto achive ions comin
‘ion of numerous activites directed at a single forsee and invariant end,
Tmevery organization, thesuccesso ts operation dependsontheco-operationThe Moral Hazard Problem
ofalits members, but the commitments are not oreseable in thelong run,
‘oraretheyinvarantor uniform In this type of organization there cannot be
Any precise delimitation of esponsibility and ite more convenient ose
‘methods that strengthen group work, Arce ia which the advantages of =
‘operation and its consequences on optimal contracts are analysed are
olmseom and Milgrom (1980), toh (1990, 1983), Macho-Stadler and
Perer-Castro (19910, 1993) and Ramakrishnan and Thakor (1991),
3C2 Organizational design in relationships with several agents
Apart om the design of contacts, he principal ao decides the oraniation
‘with which she wll rate This incudes such fundamental decision show
many agent to contract and how should they be structured. One aspect
-tadied nthe erature concerns the advantages of dlgotion of tasks which
analsesthe question of when tisbetter for he principal delegate ome task
‘wan agent instead of doing it hes
‘Consider a principal who contracts an agent to carry out some productive
task In order to obtain evidence abou the production, itis necessary to use
‘supervisory and conta mechanism. The probity of observing the rest
epends on the fort exerted by the supervisor We want to know fitisusea
for the principal to delegate the task of supervision to an agent called the
superior), oft isbeter for her to dither. iscvdent that forthe f=
‘uoduction ofa new member inthe hierarchy (inthe lationship) tobe prot.
ale itis necssary fr him tobe ableto perform some task that the principal
‘sincapable of doing, Tiss the case when the principal sles technical b-
itesthan the supervisor to cary out the tak oF ifthe costofdoingit herself is
higher. In these cases the principal should delegate supervision toa more
ficient agent Iisalso possible thatthe principal sjustarableto carry out he
supervision task asthe supervisor but hat there are commitment problems.
Think ofa someone doingadataclletion rey In onder to motivate hin
to ext elo the wage of 2 data coletor s normally based on how many
‘questionnaire have een correc filled in. In oder to now that the ie.
‘onnaireshave Been corec filed in or not the supervisor could vist few
ofthe households thatthe agent as he has vse, i ede to checkout the
authenticity of the answers to see the agent was sufficiently persuasive et
But the probability thatthe supervisor lars thatthe rorveys are properly oF
improperly done depends on the effort he daictestoisjo The prem of
lackof commitment spose inthe ellowingwayits one thing forthe super
visor to what hes going odo and anotheriwhat he reallyends up doing
‘Wermust bear ia mind tha the task of supervision isnot useful once the
2
‘The Morel Hazerd Problem
productive effort ha ben carried out, since the objective of supervision i
Preily to motivate thisefor. ence, onc the initial ob hasbeen done, he
Incentives to supervise tare minimal
‘We ned to compare the two pose sitationsthat in which the principal
takes are of the supersion tsk personally and tat in which she contract
Supeviosto do eI dhe principal and the potential person are tent,
the choice ery When the principal supervise, gen hat she cannot com
rit toher control strategy (hat toa, othe effort that she wll use and the
report that she will sen), the contrat withthe agent wll simultaneously
‘Setermine the control evel hat she willbe interested in maintaining Besides,
‘shew nly report her Gindngs iby doing so she reduces the pao that she
‘must give the agent (ithe pay-off when quality isnot observed isles than
‘when high quality sobserved the principal wll ever reporthigh quality even
iftinoberved) the principal contacteasuperviaon she wlldesign eco
teat in nde or him toe etrested in exerting the adequate level of efor,
and honest to announce what he observes, Comparing both situations, we
‘conclude that delegation ofthe supervisory task eas toasty beter sa
tion than not delegating when the principal cannot commit to acontol strat
‘egy oF the honesty a her report. Does his mean that delegating the conte
{uns alays beneficial when the principal as problems to make her behav
‘ourbeicwable Theanswerisno. Theexample we have presentedisvey simpli
fed To get more profound vision ofthe advantages and disadvantages of
You might also like