UC Berkeley Economics 209A: Theory and Application of Non-cooperative Games, Part II, Spring 2003 (co-taught with Joel Sobel; go to his course page)
The Problem Set and Exam answer keys are posted right below the Problem Set and Exam. The Problem Set answer key now corrects an error in the answer to question 1(f) and a typo in the answer to question 2(d) (where the 650 should have been 750.) Comments and questions welcome.
Judi Chan will have your graded Problem Sets and Final Exams probably in the week of April 14. Joel Sobel and I will aggregate and report your grades soon thereafter.
Office hours: Mondays, March 3, 10, 17, and 31 only, 2:45-3:45 or by appointment, at a room in Evans to be announced
Lectures: Mondays, March 3, 10, 17, and 31, 10:00-11:50 in 639 Evans and then from 12:00-1:00 in 47 Evans (just like Part I)
Organization: The only requirements for Economics 209A, Part II, are a problem set posted below, to be done independently and due at the start of class March 31 (the answers will be posted below on April 1); and a final exam on Monday, April 7, from 10:00-11:50 in 639 Evans. The problem set will count as 35% of your grade, and the exam will count 65%. Your grade on Part II will be combined with your grade from Part I, taught by Joel Sobel, to determine your course grade.
The final exam includes a half-hour essay question, which is meant to get you thinking about how to use behavioral game theory to do economics; a choice gives you some freedom to make the question about the kind of economics you are interested in. This essay question is now posted below; procrastinators and preproperators alike are strongly encouraged to start thinking about it before April 7.
Course materials (download free Adobe Acrobat Reader for pdf files)
Colin Camerer, Behavioral Game Theory: Experiments on Strategic Interaction, Princeton, 2003, a source of many of the readings, has arrived and should be available in your campus bookstore. If you have trouble finding it, you can call California-Princeton Fulfillment Services directly at 1-800-777-4726.
1. With regard to the first question in the problem set, you were not meant specifically to assume a distribution of types, but rather to use your analyses of the games and your intuitions about real strategic behavior to make predictions about what would happen in the various cases.
2. In parts b and d, thinking about behavior in terms of types may well be unhelpful. Here you are meant to think of what kinds of conventions are likely to govern real behavior in the equilibrium selection problems the questions refer to.
3. Here the answer is like that to your question on 2: Think about what real people would do in the situation, and use types only to the extent you find them helpful. (As I was suggesting in class, the natural definitions of types are somewhat different here than in non-signaling games, starting from truthfulness or credulity rather than uniform priors; but you should be able to make progress on the question even without types.)
With regard to your more fundamental question, it's not always plausible to assume such simple beliefs for L1, but it turns out to be a useful way to think about observed behavior in experiments.
At 10:18 AM 3/21/2003 -0800, you wrote:
Dear professor Crawford,
I have a few clarification questions rearding the first question in the problem set:
1. In general, should we assume in this question some distribution of the different types (Naive, L1, L2)?
2. In the extensive form games with complete information presented in this
question (e.g. parts b and d), it seems to me that we cannot apply
the types method, because when there is full observation prediction is irrelevant.
Is that correct?
3. In part e, when we talk about Naive, L1, L2 sender and receiver (of
messages), I'm not sure what's the right way to interpret those types in
this context. For wxample, is a naive receiver someone who believes the sender
and best respond or does he just play his strategies with uniform
distribution. Similarly, does the naive sender just tell the truth, or does
he tell all strategies with equal probabilities?
In case I'm not on the right track, can you please give some more guidance regarding your expectations for this question?
In addition, I have a more fundamental question about the types. Consider
for example type L1. He best responds to naive. This is like
saying that he predicts that 100% of the population (except for himself) is
naive. Is this a realistic assumption given that he, himself, play
L1?
Second email:
Yes, imagine that there are two rival notions, and try to figure out which equilibria are most likely to describe the behavior of people who are paired to play the game.
At 01:29 AM 3/25/2003 -0800, you wrote:
Prof Crawford,
I have a question regarding part (d) of question 3. I'm not sure
what exactly you mean by suppose there are two plausible, but rival, notions
of
dividing the dollar fairly. Are we supposed to present two notions and
redo the analysis based on our two notion of fair?
Copyright © Vincent P. Crawford, 2003. All federal and state copyrights
reserved for all original material presented in this course through any medium,
including lecture or print.