L5: Public good games

Question 1

Explain how the PGG works?

Answer 1

Generalised prisoners dilemma games, subjects play in group of size ‘n’
Each subject endowed E tokens, which they divide between a private and public account
Each unit in private account of ‘i’, earns one token for subject i
Each unit in public account is multiplied by k and divided by n tf return (k/n) per subject
Subjects choose independently and simultaneously

Question 2

How should self-interested parties act in PGG?

Answer 2

Should contribute nothing

If whole group does this each receives E tokens

Question 3

How much would be earned if all contributed all in PGG?

Answer 3

n.k.E tf more than before

Question 4

What are the 4 stylised findings for the standard PGG?

Answer 4

1) In 1-period PGG avg. contribution is 40-60%
2) Higher values of k -> more contributions
3) Pre-play communication has strong positive effect on contribution levels

Question 5

What are the 2 stylised findings for the repeat interaction PGG?

Answer 5

1) Initially contributions as high as in single period games

2) Contributions gradually decline over time -> full defection in final rounds (Keser and van Winden (1996))

Question 6

What does the learning hypothesis look at?

Answer 6

Why do subjects contribute?

Is it because they misunderstand the rational solution and then LEARN to free ride over time???

Question 7

Explain Andreoni’s (1988) learning hypothesis experiment?

Answer 7

Subjects do 10 rounds of PGG, then a surprise restart is introduced at the end of the 10 rounds

Question 8

What did Andreoni (1988) hypothesise?

Answer 8

If decay in contributions is due to LEARNING, the restart should have no effect on the decay of contributions

Question 9

What did Andreoni (1988) find?

Answer 9

After 10 rounds of decay, the contributions jumped up in both strangers and partners treatment (contradicts learning hypothesis)

Question 10

What was Keser’s (1996) theory?

Answer 10

That people co-operate because they make errors

Since the Nash eq. is at zero, there cannot be negative contributions, therefore all errors must be >0
Therefore, any errors would inevitable lead to contribution tf co-op and errors are indistinguishable

Question 11

How does Keser (1996) test his theory?

Answer 11

Conducts a non-linear PGG with the NE as an interior solution! (makes optimal strategy to keep 13 and contribute 7 tokens!)

Question 12

How would Keser know if his prediction of errors rather than co-operation is true?

Answer 12

True errors would fluctuate unsystematically about the NE

Question 13

What did Keser find?

Answer 13

Avg. contribution is 10.29 tokens, tf a 25% over-contribution rate - implying co-operation not error

Question 14

What is the conditional co-operation hypothesis?

Answer 14

That groups are made up of two types of people: self-interested subjects and conditional co-operators

Conditional co-operators start out by contributing to the public account, but over time they begin to realise that not everyone in the group is like them. In response, they reduce their contributions, leading to the decaying pattern.

Question 15

Explain the structure of Fischbacher et al. (2001) and what they were testing?

Answer 15

Testing conditional co-operators hypothesis:

1) Unconditional co-op.: participants have to decide how much to contribute to the public account
2) Next, are asked how much they are willing to contribute for each one of the possible avg. contributions of the other group members

Question 16

What did Fischbacher et al. (2001) find?

Answer 16

Conditional co-operators start out by contributing to the public account, but over time they begin to realise that not everyone in the group is like them. In response, they reduce their contributions, leading to the decaying pattern.