Docstoc

Intrinsic and Extrinsic Motivation.ppt

Document Sample
Intrinsic and Extrinsic Motivation.ppt Powered By Docstoc
					Intrinsic and Extrinsic Motivation

          Robert Östling
                    Outline
•   Experimental results
•   Theoretical considerations
•   Bénabou & Tirole (2003, RES)
•   Bénabou & Tirole (2006, AER)
•   Ellingsen & Johannesson (2006)
•   Herold (2005)
    Frey and Oberholzer-Gee (1997, AER)

•   Question: Tests reaction to monetary compensation for the
    acceptance of a nuclear waste repository in their community.
•   Setting: Hypothetical referendum voting question was posed to
    subjects about if they were willing to permit the construction of a
    nuclear waste repository in their community. The same question was
    then repeated with a specific monetary compensation for accepting the
    nuclear waste repository (varied between $2175, $4,350 and $6,525
    per individual and year in three sub-samples). Everyone who rejected
    the compensation then received another question with a higher
    compensation ($3,263, $6,525, or $8,700).
•   Subjects: 305 interviews of randomly selected subjects in two
    communities in Switzerland (potential sites for a nuclear waste
    repository).
•   Results: Fraction of acceptance without monetary compensation:
    50.8%. Fraction of acceptance with monetary compensation: 24.6%
    (the amount of compensation had no significant effect on the
    acceptance probability).
    Gneezy & Rustichini (2000, QJE)
•   Setting: Subjects were asked to answer a quiz consisting of 50 problems taken out of a
    psychometric test used to scan university applicants (questions involving reasoning and
    computation).
•   Treatments: 0 NIS (except show-up), 0.10 NIS, 1 NIS or 3 NIS per correct answer
•   Subjects: 160 subjects
    Gneezy & Rustichini (2000, QJE)
•   Setting: High-school students collected donations from the public to some charity (done in
    Israel a few days a year). Students divided into pairs of students who collect money
    together as a team.
•   Treatments: Baseline (speech), baseline + 1% of collected money, baseline + 10% of
    collected money.
•   Subjects: 180 Israeli high-school students.
    Gneezy & Rustichini (2000, JLS)
•   Setting: A study of day-care centers in Israel. The day care centers operates between 7.30 and 16.00.
    Before the study there was no fine if parents came late to pick up their children.
•   Treatments: Control (only record late parents) and treatment (recorded first 4 weeks, then a fine of 10
    NIS for late pick-up, removed fine in 17th week).
•   Subjects: The study was carried out on 10 day-care centers in Israel (center 1-6 in the test group and
    center 7-10 in the control group). Between 28 and 37 children in each day care center.
         Fehr & Rockenbach (2003,
                  Nature)
•   Setting: A trust game is played anonymously between two individuals (both endowed with
    10 MU). The amount sent is tripled. Investors always state desired back-transfer.
•   Treatments: Trust condition (no fine possible) and incentive condition (4 MU fine if less
    than stated back-transfer).
•   Subjects: 138 subjects.
    Mellström & Johannesson (2005)
•   Setting: Subjects are given the opportunity to become blood donors (complete a health examination
    consisting of a health declaration and a physical examination at a blood donation center). Subjects are
    offered different incentive schemes to become a blood donor depending on the experimental treatment.
•   Treatments: No compensation, 50 SEK compensation, a choice between SEK 50 or donating SEK 50 to
    charity.
•   Subjects: 262 subjects.
     Falk & Kosfeld (2006, AER)
•   Setting: Principal has zero endowment, agent has 120 MU endowment. Agent choose how much to
    transfer to principal knowing that the transer is doubled. The principal has choice whether to restrict the
    agents choice set or not. Used strategy method.
•   Treatments: C5: Principal can impose restriction of at lest 5 MU transfer or choose not to do so. C10:
    Same but 10 MU as minimum. C20: Same but 20 MU as minimum.
•   Subjects: 418 subjects.
     Theoretical considerations
• Extrinsic and intrinsic motivation (e.g. Frey and
  Oberholzer-Gee 1997): Individuals have an intrinsic value of
  performing some activities (e.g. helping others, civic duty) and
  external monetary compensation reduce the intrinsic value.
• Incomplete contracts (e.g. Gneezy and Rustichini 2000): The
  introduction of a monetary payment changes the perception of
  the incomplete contract (the introduction of a fine for coming
  late changes the perception of the consequences of frequent
  delays).
• Reciprocity (e.g. Fehr & Rockenbach 2003): The introduction
  of a fine is viewed as an hostile/unfair act, and is therefore
  punished by reduced generosity.
• Signaling: The decision to introduce a fine or incentive signals
  some private information on part of the principal. Can
  incorporate all of the above explanations.
          ?                                                                     ?
                                            b


                                            e


B&T ’03   Private information about A (or   Imperfect self-knowledge (from private signal): ”I’m not
          task).                            trusted, so probably I am not trustworthy and won’t
                                            work so hard.”
B&T ’06   No private information.           Private information about own prosociality and greed
                                            with reputational concerns: ”The principal trusted me so
                                            now I can really show off that I am trustworthy rather
                                            than greedy.”
E&J ’06   Private information about own     Private information about own characteristics and
          expectations about A (resulting   esteem concerns: ”The principal is not trusting so I
          from private information about    don’t need to work so hard to live up to expectations.”
          own characteristics).
H ’05     Private signal about A’s type.    Private information about own characteristics
                                            (trustworthy/untrustworthy) and uncontractible part of
                                            relationship: ”He didn’t trust me, so there is no point in
                                            doing something extra outside our contract.”
Bénabou & Tirole (2003, RES)
• A choose e=0 or e=1 at a cost c. Task succeeds with
  probability θ if undertaken.
• P knows c, whereas A only gets a signal σ that is
  informative about c.
• If task succeeds, A gets V>0 and P gets W>0. If it
  fails, both get zero.
• P chooses whether to reward effort with a bonus
  b≤W or not.
• If A knew c, he would exert effort only if θ(V+b)≥c.
• With two-sided private information we look for a PBE
  of the two-stage game.
Bénabou & Tirole (2003, RES)
• Reminder: A PBE is a strategy profile and
  belief system such that strategies are
  sequentially rational at each node of the
  game tree and beliefs are consistent.
• At the second stage A forms an expectation
  E(c|b, σ) and exerts effort only if θ(V+b)≥
  E(c|b, σ) which implies that there is a
  threshold σ* for the agent to exert effort.
• P choose b in order to maximize expected
  payoff: Eσ[UP]= θ[1-G(σ*(b)|c))[W-b].
• Ignoring degenerate equilibria.
     Proposition 1 (B&T ’03)
• Bonus is a short term reinforcer: if b2>b1,
  then σ*(b1)>σ*(b2).
• Bonus is bad news, if c2>c1 then b2≥b1.
• Bonus undermine A’s perception about task
  attractivness: if b1<b2, then E(c|b1, σ) <
  E(c|b2, σ).
• Conclusion: Extrinsic motivation can crowd
  out intrinsic motivation in the long term.
      Bénabou & Tirole (2006, AER)
     • P has no private information, only A knows
       his identity (ve and vb). A engages in social
       signaling.
     • Agent effort level e to maximize the
       following:
     (ve+vbb)e – C(e) + µeE(ve|e,b) – µbE(vb|e,b).

  Intrinsic      Intrinsic   Material    Reputational     Reputational
 immaterial      material    cost of       effect of        effect of
  valuation      valuation    effort      observer’s       observer’s
(prosociality)    (greed)               expectation of   expectation of
                                         prosociality        greed
Bénabou & Tirole (2006, AER)
The first order condition is:

C’(e) = ve+vbb + µe∂E(ve|e,b)/∂e – µb∂E(vb|e,b)/∂e.

1.   Observing e reveals the sum of three marginal
     motivations (intrinsic, extrinsic and reputational).
2.   Higher b increases informativeness of vb and
     reduces it of ve.

Assume µe and µb are identical and known to all.
vb                                     b=0
                              +

                                                                     Potential
                                                                     crowding
                                                                     out!
              b>0 without
              reputation
                                               b>0 with
                                               reputation

                                        −

                            C’(e)-r(e,0)=ve*     C’(e)-r(e,b)=ve**
                                                                           ve
Reputational effects:
1.  New contributors have lower ve which may diminish the
    reputational effect of e.
2.  New contributors are greedy, i.e. have higher vb, which
    reduces the reputational effect of e.
Bénabou & Tirole (2006, AER)
• Interpretation of Mellström & Johannesson
  (2005): Women may not be more prosocial,
  but care more about appearing prosocial!
• Argument does not rely on private
  information of P, but requries
  multidimensional characteristics.
• Implies that P can be replaced by computer
  or experimenter (blood donation okay, but
  what about workplace relationships?)
 Pride and Prejudice (E&J ’06)
• P private information about expectation of A’s type.
• A private information about own type, but cares
  intrinsically about P’s expectation (esteem).
  Depending on the situation, esteem from P might
  depend on P’s type (salience weights).
• Requires heterogeneity among both P and A.
• Heterogeneity of P’s expectations violates the
  assumption of common priors.
• So first let us derive these beliefs from common
  “metaprior” in a simple two-type case.
    Agreeing to Disagree (E&J ’06)
•   Two types: θH>θL
•   Two distributions: P(θH|G)=h>P(θL|B)=l.
•   Metaprior of θH: h/(h+l)
•   Prior for distribution of θH type:
          P(G|θH)=h/(h+l)>1/2
•   Prior for distribution of θL type:
          P(G|θL) = (1−h) / (2−h−l)<1/2
•   Calculating the prior for θH given θH and θL shows that a high
    type attaches a higher probability to θH than a low type.
•   Common metapriors and type heterogeneity delivers different
    priors
•   A high type is optimistic, a low type is pessimistic.
    Ellingsen & Johannesson (2006)
•   Crowding out: High-powered incentives may signal low expectations, implying
    that A does not need to work so hard to get P’s esteem. This argument requires
    that esteem is worth less from a pessimistic, low type, P.
•   Gift exchange: If esteem from a high type, optimistic P is worth more to A,
    then P may signal he is a high type by paying a higher wage and A will work
    harder to get the esteem benefit
•   Hidden cost of control: Exogenous imposition does not matter (since it does
    not carry expectations). However, a controlling P is a relief for A since esteem
    becomes cheaper – asymmetric information may consist of different beliefs
    about what is appropriate behavior in a situation.
•   Social salience channel: Similar effects might arise from different views of
    what appropriate behavior is. P’s actions might signal what he believes is
    appropriate and A will be interested in this because it signals how to achieve
    esteem. The above mechanisms can be restated in this way, but it more
    generally implies that esteem-seeking agents are sensitive to cues about
    appropriate behavior. May explain how different types of rewards (e.g. small
    bonus,gold star) can be important, but more generally about framing (PD as
    Wall Street Game vs Community Game).
Example: Trust Game (E&J ’06)
  Example: Trust Game (E&J ’06)
• Relevant characteristic: Fehr-Schmidt inequality
  aversion.
• One equilibrium for “reasonable” parameters: Only
  high type P1 are trusting, and high type P2 rewards
  trust only in the voluntary trust game.
• Idea: Esteem more valuable from a high type P1 and
  it is only in the voluntary trust game that P2 can be
  sure that P1 is a high type.
• Key assumption: Low type esteem for P2 be
  sufficiently low compared to high type esteem for P2.
         Into the future…
• Relation to psychological game theory
• Bounded rationality (e.g. Bayesian
  updating)
• More than two players (too complex?)
                            Bye, bye!
• Framing, social preferences, symbolic
  interactionism, meaning et c.

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:54
posted:5/7/2012
language:English
pages:25
zhaonedx zhaonedx http://
About