Folie 1 - Uni Kassel

Experimental Economics:
Negotiations & Real Effort Experiments
Prof. Dr. Björn Frank
Winter Term 2013/2014
1
How to obtain credits:
You prepare and perform an experiment (group work)
You present the experiment's result (plus report of 4 pages
or so, all group work)
You write a seminar paper on one of the seminar topics
(except that of "your" experiment). 10 to 12 pages, no
group work.
In case you only need 3 credits, requirements can be negotiated.
2
Topics for presentations and/or papers
1. What difference does real effort make? The
dictator game
2. Who likes to compete?
3. Framing the Endowment Effect
4. Ultimatum games with real effort
5. The role of time in (ultimatum) negotiations
6. Communication in negotiations
7. Experimental investigation of Coasian
negotiations
4
Schedule
23.10.
30.10.
6.11.
13.11.
20.11.
27.11.
4.12.
11.12.
18.12.
15.1.
22.1.
29.1.
5.2.
12.2.
Topic allocation
Lecture / Experiment Bühren&Khachatryan
Lecture
Experiment 1. What difference does real effort make? The dictator game
Experiment 2. Who likes to compete?
Presentation
Experiment 3. Framing the Endowment Effect
Experiment 4. Ultimatum games with real effort
Presentations
Experiment 5. The role of time in (ultimatum) negotiations
Experiment 6. Communication in negotiations
Presentations
Experiment 7. Experimental investigation of Coasian negotiations
Presentation
5
In economic experiments, subjects are
confronted with clearly defined (often
strategic) decision problems, and their
decisions are relevant for their actual
monetary payoffs.
Control
Field
Lab
Laboratory Experiment
Charitable Giving as a
Gift Exchange
Classroom
Experiment
Regional minimum
wage schemes
Macroeconomic data
Penicillin
0
What is "control"?
• Subjects are randomly allocated to "control
group" and "experimental group" (random
assignment)
• The difference between control group and
experimental group has been designed by the
experimenter (treatment variable)
• Examples for treatment variables:
- Payment high/low
- Group decision versus individual decision
- Communication allowed yes/no
- etc.
Changing two variables at once (in
comparison to control group), you don't
know what caused the effect
c
a
b
d
ac
bd
Solution:
Have observations for all four quadrants of the matrix
(2x2 between subjects factorial design)
Besides control, another important reason
for doing experiments is simply that some
phenomena cannot be explored with real
world data
Why do experiments?
(1)-(3): Roth (1995) plus Vernon Smith (1994)
["Economics in the Laboratory", Journal of
Economic Perspectives 8, 113-31]
(1) speaking to theorists
- Test a theory or discriminate between theories,
- Explore the reasons for a theory's failure,
(2) searching for facts
- Establish empirical regularities as a basis for new
theory
- Compare environments / institutions
(3) whispering in the ears of princes
- Evaluate policy proposals
- Use the laboratory as a testing ground for
institutional design
(4) replication
- Checking the (internal) validity of previous studies
Why do experiments?
Normative reasons
Positive Reasons
(1) speaking to theorists
- Finding support for own theory
(2) searching for facts
- It's fun
- "Automatic" fulfillment of
requirements for doctoral thesis
(3) whispering in the
ears of princes
- "We are important"
(4) replication
?
Experiments v. Simulations
• "real" choices and "real" decisions
independent of researcher
• computer simulations: extension of
researcher's brain, not "real" data like
experiments
External validity
• "Results obtained in experiments do not necessarily
allow us to infer on actual behavior of economic
agents." Heinz Sauermann
•
"Ergebnisse, die aus Experimenten gewonnen werden,
lassen nicht unbedingt einen Schluß auf das tatsächliche
Verhalten von Wirtschaftssubjekten zu." Heinz Sauermann
• TGN1412 disaster
(Javaneraffe)
Cynomolgus monkey
Solution in economics
(not in medicine!):
Field experiments
Why
External Validity might be sometimes low
• Observation (experimenter effects)
• Context / Framing
• Incentives
• Self selection
- Special group of those who take part in
experiments
- Special group of real world decision makers
List & Levitt 2005
experimenter effects
The Ultimatum Game (Güth, Schmittberger und Schwarze 1982)
Proposer
x
0
Responder
100
accept
reject
100 – x
x
0
0
experimenter effects
in the ultimatum game
Frank (1998):
- stake size stamps worth DM 5
- 20 pairs control group, [22 pairs experimental group
where stamps are burnt in case of rejection]
- average offer DM 2.06 [1.96]
- average "minimum acceptable offer" of responders:
DM 1.32 [1.12]
- Offer that maximizes expected payoff:
DM 2.00 [2.00]
See Zizzo (2010) for a survey of experimenter demand effects
Please decide between a and b
b
a
a
9; 9
0; 8
b
8; 0
7; 7
Framing
The let's
get 7 game
The let's
get 9 game
b
a
a
b
a
9; 9
0; 8
a
9; 9
0; 8
b
8; 0
7; 7
b
8; 0
7; 7
Ross and Ward (1995)
Cooperation rate in prisoners' dilemma…
-
When it is called "Community Game": 66%
-
When it is called "Wall Street Game": 33%
Evidenz on external Validity
• Bortolotti/Casari/Pancotto (2013), Norms of
Punishment in the General Population, mimeo:
Showing that results obtained with students cannot be
generalized to the society at large
• But see Exadaktylos/Espin/Branas-Garza (2012),
Experimental Subjects are Not Different, mimeo, for a
different result (in dictator, ultimatum and trust games)
• Guillen/Veszteg (2012), On "lab rats", Journal of SocioEconomics 41, 714-720, investigate the determinants
of the decision to return to the laboratory and find that
men and subjects with relatively high payoffs are more
likely to participate again.
Evidenz on external Validity
• Ernst Fehr & Andreas Leibbrandt: Cooperativeness and
Impatience in the Tragedy of the Commons, IZA DP
No. 3625 August 2008
• Experiments with fishermen at a lake in the Northeast
of Brazil
• PGE and TPE (1 bottle of mineral water immediately or
2 on the next day)
Blair L. Cleave, Nikos Nikiforakis and Robert Slonim (2010), Is There Selection
Bias in Laboratory Experiments?, mimeo
"In our experiment Player A received a $20 endowment and chose
whether to send $0, $10 or $20 to Player B. The amount Player A sent
was tripled and given to Player B. Player B chose how much to return to
Player A from the amount he received." = TRUST GAME
Aus
Cleave,
Nikiforakis
und
Slonim
(2010)
Instructions
- Written
- Typically not loaded
- Avoid "ad-libbing" (improvisation)
- Avoid deception
Deception
"Most economists are very concerned about developing and
maintaining a reputation . . . for honesty in order to ensure
that subject actions are motivated by the . . . monetary
rewards rather than by psychological reaction to suspected
manipulation. Subjects may suspect deception if it is
present. Moreover, even if subjects fail to detect deception
within a session, it will jeopardize future experiments if the
subjects ever find out they were deceived and report this
information to their friends'' Davis/Holt, 1992, pp. 23, 24
"It is believed by many undergraduates that psychologists
are intentionally deceptive in most experiments. If
undergraduates believe the same about economists, we
have lost control." Ledyard, 1995, p. 134
Deception
"(I)t is crucially important that economics experiments
actually do what they say they do and that subjects believe
this. I would not like to see experiments in economics
degenerate to the state witnessed in some areas of
experimental psychology where it is common knowledge
that the experimenters say one thing and do another. This
would be very harmful to experimental economics. (...)
[Subjects] believing what the experimenters tell them . . .
seems to me to be of paramount importance: once subjects
start to distrust the experimenter, then the tight control that
is needed is lost.'' (Hey, 1991, p. 171, 173)
Repeated trials
- "How much attention should we pay to
experiments that tell us how inexperienced
people behave when placed in situations with
which they are unfamiliar, and in which the
incentives for thinking things through carefully
are negligible (...)?" Binmore 1994
- "When major decisions are involved, most
people get too few trials to receive much
training" Thaler 1987
- Often training sessions and test questions to
check subjects' understanding
I - IV: Nagel (1995)
V: Game Theory lecture in Kassel, 2008
Period
1
2
3
4
V
I
II
40,7
29,8
30,4
26,6
39,7
28,6
20,2
16,7
37,7 32,9 36,4
20,2 20,3 26,5
10,0 16,7 16,7
3,2
8,3
8,7
III
IV
Nagel, Rosemarie: "Unraveling in Guessing Games: An Experimental Study," American
Economic Review 85 (1995), 1313-26
Frohlich and Oppenheimer (1998) let groups of 5 people play a
15 rounds multilateral prisoner's dilemma without any previous
communication, with e-mail communication before each of the
first 8 rounds, or with face-to-face communication before each
of the first 8 rounds. In every round, each player can give any
amount between 0 and 10 to the group, keeping the rest. The
sum of all contributions is multiplied by 0.4, and each player
receives the resulting amount.
(If everyone contributes 0, everyone gets, or rather keeps, 10 per
round. If everyone gives 10, everyone gets 20 per round.)
Isaac and Walker (1988):
Group size either equals
4 or 10
H denotes MPCR = 0.75
L denotes MPCR = 0.3
Financial incentives
- Usual practice: Let the expected payoff of the
subjects' equal their typical hourly wage
- Possible, though not used for the majority of
experiments: random pay
- Random round payoff mechanism: E.g., play
four experiments, and afterwards randomly
determine for which you actually pay
- Random person payoff mechanism: Randomly
determine which participants are actually getting
a (high!) payoff
The Ultimatum Game
What difference does a larger cake size make?
Hoffman et al. 1996: Stake size 100 US-$, two out of 6
responders rejected 30 $
Cameron (1999): Stake size 5000, 40,000, 200,000 Rupiah,
the latter is about a month' wage in Indonesia.
Average offer: 42% / 45% / 42 %, hence no impact of stake
size.
Rejection rate slightly lower for higher stakes.
But see Andersen/Ertaç/Gneezy/Hoffman/List (2011), Stakes
Matter in Ultimatum Games, AER 101, 3427-3439, with
evidence from India
Becker deGroot Marschak
• How to get an unbiased willingness to pay (or willingness
to accept)
• Example: Güth/Weck-Hannemann: Do people care about
democracy? An experiment exploring the value of voting
rights, Public Choice 91 (1997), 27–47
• Buying price for voting right: Between 1 and 200 DM
(Actual price 112 DM)
Postexperimental questionnaire
Strategy method
(strategy elicitation method)
• Selten 1967
• Example: Ultimatum game
• Possible problem: Impact on subjects'
behavior (Brosig/Weimann/Yang: The Hot Versus
Cold Effect in a Simple Bargaining Experiment,
Experimental Economics, 6 (2003), 75–90)
Jordi Brandts and Gary Charness: The Strategy
versus the Direct-response Method: A Survey of
Experimental Comparisons, Experimental Economics
14 (2011), 375–398
James Andreoni, Marco Castillo und
Ragan Petrie (2003), What Do
Bargainers' Preferences Look Like?
Experiments with a Convex.
Ultimatum Game, American
Economic Review 93, 672-685
41
Brosig/Weimann/Yang 2003 p.77
"B" is punished by playing "1" in 42%
of all cases in which "B" was played, but
only in "hot"-mode (cold: 4%)
Brosig/Weimann/Yang 2003 p.77
(Inefficient)
NGG
43
NEP Cognitive and Behavioral Economics
Camerer, Behavioral Game Theory, Princeton U Pr 2003 - sehr schön
aufgebaut für Leute mit leichten Vorkenntnissen in Spieltheorie
Friedman/Cassar: Economics Lab, Routledge 2004 - kurz und bündig
mit praktischen Tips
Martin Held et al (Hrsg, 2003), Normative und institutionelle
Grundfragen der Ökonomik Jahrbuch 2: Experimente in der Ökonomik. Nett zum Schmökern
Kagel/Roth (ed.), Handbook of Experimental Economics, Princeton
University Press 1995
Friedman/Sunder, Experimental Methods, Cambridge UPr. - "Kochbuch",
aber für meinen Geschmack zu marktexperimentlastig
Davis/Holt, Experimental Economics, Princeton UPr 1993
Plott/Smith (ed.), Handbook of Experimental Economics Results, NorthHolland 2008 - Nagelneuer Wälzer mit über 1000 Seiten, zu
unübersichtlich und heterogen für einen Einstieg in dieses Feld