Transcript Bringing Economic Experiments to the Classroom

FEELE Lab
Finance and Economic Experimental
Laboratory at Exeter

Todd Kaplan

Dieter Balkenborg

Tim Miller
FDTL5 Grant for Bringing
Experimental Economics
into the Classroom
Bringing Economic
Experiments into the Classroom
Objective: Put experiments into the undergraduate
economics curriculum at the University of Exeter.
Thanks to:
Higher Education Academy in England for the £s.
Economics Network (England)
Dept. of Economics at Exeter.
Steve Gjerstad, Denise Hazlett, Charlie Holt
Why use classroom experiments?

Inspires students.






Chamberlin 1948 ran a simple market experiment.
This inspired Vernon Smith to go into experimental
economics.
Todd Kaplan was a student at Caltech 20 years ago.
Experiments with Charlie Plott inspired him to switch majors
to econ.
Memorable experience for students.
Brings students closer to research. Allows research to
help teaching and teaching to help research.
We have embedded 48 experiments into 18
undergraduate modules at Exeter.
Types of Classroom Experiments
HandRun.




Quick raise hands (symmetric prisoners’ dilemma).
Sampling paper collection (2*2 games, currency attack).
More sophisticated (Pit market, Sloman’s Trade Game).
Computerized.



Web based: PHP or Java (Bertrand, Double Auction).
Locally based/installed (z-tree).
Homework.





Simple Q&A with feedback in class: (Rubinstein’s site).
More advanced Individual Choice experiments with some immediate
feedback (Monty Hall).
Play against a fictitious/robot/prior human player (Holt: Traveler’s
dilemma).
Students play each other at designated time.
Research.




Single lecture (Chamberlain)
Complete semester (Selten/Mitzkewitz/Uhlich, Iowa Pol. Stock Market)
Requirement to be a subject. (Psychology)
Hand run


Advantages:
 Suitable for large lectures.
 Some take just minutes.
 Engaging for students .
Disadvantages:
 May require careful preparation, including room structure.
 May require assistants/volunteers.
 May require lecturer w/ practice: Student experience may vary.
 Data collection/entry takes time. Feedback may be delayed. (the
next lecture)
 One can only run for few rounds.
 Some students may not be participating.
Handrun Hints:

Techniques for low setup costs.







At beginning of semester, cut simple strips of paper (you may want to
use two colours).
Bring several plastic bags from home to collect answers.
Give verbal instructions/ display question.
Quick to collect and sample a few.
Display last year’s results.
Know what you are doing beforehand, particularly with
assistants.
Example: Guessing game.



Guess a number 0 to 100.
The guess closest to 2/3 the average number wins a prize.
Ties will be broken randomly.
Graph of guesses (Exeter UGs)
Average was 36.7 (winning guess 24.5), equilibrium is 0!
18
16
14
12
10
8
6
4
2
0
0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90
Six guesses were above 66.66
90100
Exeter UG 2nd round guesses
Average was 12.4 (winning guess 8.3)
45
40
35
30
25
20
15
10
5
0
0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90
90100
Todd’s quote “you shouldn’t ignore theory, even
Oneitguess
was above 66.66
when
is wrong.”
Computerized

Advantages:



Readily available, great for beginners, uniform
experience.
Immediate results, data ready for evaluation
Disadvantages:





Experiments standardized, limited flexibility
Room requirements, split large lectures, use tutorials
May need trained teaching assistant
Equipment (PhP vs Java, handhelds)
Needs one hour of teaching time
Computer Experiment Hints




Try using two students per computer (cuts down on
web surfing, email browsing and gossip).
Give instructions beforehand (helps foreign & dyslexic
students).
Let students play all treatments in easiest order. (within
subject treatments)
Example: Bertrand Competition.



Students act as firms in a market.
They choose prices each period.
Customers go to the firm with the lowest price.
Sample result: Bertrand Game
Average Price
Average Selling Price
8
7
6
Price
5
4
Marginal Cost
3
2
1
Two Firms
Two Firms
Five Firms
Fixed Partners
Random Partners
Random Partners
3
11
21
0
1
5
7
9
13
15
Time
17
19
23
25
27
29
“I “Some
learnt that
collusion
can take place
in a Seriously
people
are undercutting
bastards!!!
competitive
even without
actual
though, it market
was interesting
to seeany
how
the theory is
meeting
place between the two parties.”
shown taking
in practise.”
Homework

Advantages:





Disadvantages:



Saves lecture time
Less hassle
Many periods possible (no time limit)
Easy access to homework data
Typically not interactive
Participation rate can be low unless incentives are in
place.
Hint: use some sort of incentive.
Research

Advantages:



RESEARCH
Low cost (saves recruitment costs).
Disadvantages:


Organization needs to be more careful than standard
teaching experiments.
Limited communication between and to students.
Challenges for Classroom
Experiments



Students
Lecturers
Module (Course) Structure
Challenges
Students:
 They have limited time too (both inside and outside the
classroom).
 Maintain attention (negative externality of both data
and speed).
 Assess performance/motivation (earnings is noisy,
attendance has problems).
 How can we write test (or homework) questions that
require student to attend experiment?
 Some may (correctly/incorrectly) resent being used for
research experiments at a sacrifice of their studies.
Challenges
Lecturers:
4 types: Eager, Interested but need support,
hesitant, unwilling.
 Need to want to try something new.
 Need to sacrifice course time.
 Need to think experiments are of value.
Challenges
Classes:
 Many courses don’t have experiments that fit
directly into them (macro / finance).
 Same experiments for different courses (avoid
repeat).
 Some experiments may require students to be in
two courses. (Run an experiment in micro
economics and analyze it for the statistics
course.)
Rewards
Students
• Enjoyable, Interactive
• Better grades (Emerson & Taylor, 2004)
• Concrete learning experience (rather than
abstract/mathematical formulas).
• Some students do well in experiments but poor on tests
(such as those with poor math skills).
Lecturers
• Richer teaching style -> evaluations
• Promote experimental economics (preach what we practice).
How to use experiments
Size can determine how:
 Large lectures (>100): use short hand-run, homework
 Medium Lectures (40<#<100): make use of computerized
experiments in tutorials.
 Small Lectures (<40). Possible to use computerized
experiments in place of lectures
Experiments seem to work well for all levels of students (even high school)
Hints:
 Usually do experiments before covering the material in the
course.
 Let students participate in preparation, execution and
evaluation. (Especially in an experimental class.)
 Relate some exam questions to experiments.
 Don’t be too obsessed with preserving a research
environment.
Exeter Games




On purpose same style as Veconlab (Holt).
Goal is to complement selection of experiments.
14 computerized experiments.
Types are a mix of individual choice (homework)
and interactive games.
Exeter Games Experiments
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
Bertrand Market (micro / IO)
Call Option (finance, homework)
Currency attack (macro)
Diamond Dybvig bank run (macro /banking)
Insurance (economics of social policy, micro)
Holdup Problem (corporate finance)
Kiotaki Wright fiat money (macro /banking)
Lemon Game (micro / IO, homework)
Monty Hall (statistics, homework)
Network Externalities (micro / IO)
Price Discrimination (micro, homework)
Team Draft (game theory)
Warren Buffet investment (finance)
Introductory Statistics (statistics)
How to find Exeter Games



Google for “Feele” to find the FEELE Laboratory
homepage.
Ignore the large link “Register here”, which is for
subjects to register for one of our research experiments.
Instead scroll down to “Access our website of teaching
experiments”.



View list of experiments (short descriptions of each)
Experimenter access (soon to be you!)
Participant access (your students use this to log in; you use it to
check that your experiment is set up correctly)
Brief descriptions of each of our experiments.
Click here to register as an experimenter and run
your first experiment.
Start by registering your
email address.
The Code Word is ‘deadsea’.
Username will be based on
your initials.
Check your details and confirm.
Username and Password will be emailed
to you, after which you can log in here.
Enter details from registration
email and click here.
Click here to create your first experiment.
Choose a code word to
allow students to log in
to your experiment.
Your students will need to know
this Access Code to be able to log
in to your experiment.
Click here to configure, run and
monitor your new experiment.
The new experiment is empty and
you have to add the first session.
Click here to makes changes to
the default configuration.
Enter the number of students
in your class (default: 2).
Enter the number of rounds
you want to run (default: 1).
Save your changes.
Useful feature: you can view the
effects of your configuration
changes on the subject
instructions by clicking here.
Don’t forget: you must start
the experiment before the
first subject can log in!
Click here to monitor subjects as
they log in and view the results of
the experiment.
Open a NEW browser window (not a tab), locate the
FEELE homepage and click here to log in as a test
subject in your new experiment.
Your students will log in here.
The student enters his/her
names here. (Note: you
may enter “test” to
suppress the display of
subject instructions
during testing.)
Consists of your Username
(tk2), a hyphen (-) and
whatever code word you
chose when you created your
experiment (network).
The student checks his/her details and confirms.
You now need to open a further 5
NEW browser windows (not tabs)
and log in the remaining test
subjects.
This is what a typical result set looks like.
Breakdown of test
question responses.
Draws graph
Exports results to Excel
of results.
(space delimited).
Reminder of your
configuration
settings.
Reminder of
subject
instructions.
Subjects
often want
to know
who won!
How to find “our” Wikiversity site


Google for “Feele” to find the FEELE
Laboratory homepage.
Scroll down to “Wikiversity Site for Economic
Classroom Experiments” and click on “View
Wikiversity Site”.
On “our” Wikiversity site


Why use experiments in teaching economics
List of Experiments
Hand-run as well as computerized
 Not just Exeter Games




Courses
Hints for running economic classroom
experiments
Useful Links and Related Literature

Links to popular Veconlab (Holt) and Econport sites.
Quick link to Wikiversity.
Links to other experiment sites here.
Quick link to Wikiversity.
Resources

Web based resources.
Veconlab (Holt)
 Exeter Games
 Econport (Cox, Gjerstad)
 Game Theory website (Rubinstein)
 Denise Hazlett’s website
 Experonomics
