Assignment 1
Assignment 1
1. Construct an example with 4 candidates and at most 15 voters. Apply the 7 voting rules dis-
cussed in class to the example, such that each candidate is selected as the winner by at least one
voting rule.
应该有写过,别担
2.心(a) Provide one rule discussed in class that is vulnerable to the lack-of-monotonicity paradox.
Construct a counter-example.
(b) Provide one rule that is vulnerable to the no-show paradox. Construct a counter-example.
(c) Provide one rule that is immune to both the lack-of-monotonicty paradox and the no-show
paradox. Explain your answer.
3. Consider the following voting rule: Each voter is endowed with a total of 100 points; A voter
can assign an arbitrary number of points to a candidate; Each candidate’s score is the sum of points
assigned to him/her; The candidate with the highest score is selected as the winner.
(a) Is the above rule immune to the Condorcet winner paradox? Explain your answer.
(b) Is it immune to the absolute loser paradox? Explain your answer.
(c) Is it strategy-proof? Explain your answer.
4. Let us prove that there is no matching mechanism satisfying both stability and strategy-proofness.
Consider the following marriage problem.
1
Rm1 Rm2 Rm3 Rw1 Rw2 Rw3
w2 w1 w1 m1 m3 m1
w1 w3 w2 m3 m1 m3
w3 w2 w3 m2 m2 m2
For parts (c) and (d), let ϕ be a stable matching mechanism. Then, ϕ would select one of the two
stable matches.
(c) Suppose that ϕ selects the match more preferred for all the men. Woman w1 misreports her
preferences to be such that m1 is the most preferred and m3 is the least preferred.
Verify that now there is only one stable match. Is w1 better off in this new match, compared to the
initial match selected by ϕ?
(Since there is only one stable match after w1 ’s manipulation, ϕ has to select this match. If w1 is
better off, then she does benefit from misreporting her preferences.)
(d) Suppose that ϕ selects the match more preferred for all the women.
Find one man who can misreport his preferences such that (1) there is only one stable match after
his manipulation; (2) he is better off in the new stable match, compared to the initial match selected
by ϕ.
(Combining all the above results, we can conclude that no matter which stable match is selected by
ϕ for the initial problem, there exists one person who can benefit from misreporting preferences.
Therefore, ϕ is not strategy-proof. Since ϕ is an arbitrary stable matching mechanism, there is no
matching mechanism satisfying both stability and strategy-proofness.)
2
(b) Prove that if a match is stable, then it is efficient.
(c) If a match is efficient, is it necessarily stable? Explain your answer.
1 2 3 4 5 6
h2 h4 h4 h2 h1 h2
h6 h5 h6 h6 h3 h3
h3 h1 h5 h1 h5 h4
h1 h3 h2 h4 h6 h5
(a) Suppose that no one owns a house. Let the priority order be 1 ≺ 2 ≺ 3 ≺ 4 ≺ 5 ≺ 6, where 1
has the highest priority.
What is the assignment selected by the Sequential Priority mechanism associated with this priority
order?
(b) Suppose that for each i ∈ {1, 2, 3, 4, 5, 6}, person i owns house hi .
Find all the core assignments for this Housing Market problem.
(c) For each i ∈ {1, 2, 3, 4}, person i owns house hi . 5 and 6 are non-tenants. h5 and h6 are vacant
houses. Let the priority order be 6 ≺ 5 ≺ 4 ≺ 3 ≺ 2 ≺ 1, where 6 has the highest priority.
What is the assignment selected by the You-Request-My-House-I-Get-Your-Turn mechanism?
Does the YRMH-IGYT mechanism always select an efficient assignment in House Allocation with
existing tenants? Explain your answer.
3
Open Research Questions
(These questions are optional. There is no bonus or penalty. If you have some ideas, please share.)
1. Suppose that you would like to define a modified roommate model where besides a group of
people N = {1, 2, · · · , n}, there is a set of rooms H = {h1 , h2 , · · · hn }. The rooms are heterogenous.
Each room can accommodate exactly two people. Each person cares about both his/her roommate
and the assigned room.
How to define stability in this model?
Does a stable match always exist?
If so, how to design a stable mechanism?
Is there any real-life application?
2. In House Allocation with money transfer, we might be able to define an auction-based mecha-
nism that always selects an envy-free assignment. We run a sealed-bid first-price auction for each
house hj . Each person i’s bid is his utility from the house uihj .
How to deal with the cases where one person wins multiple houses?
How to achieve budget balancedness?
How to guarantee envy-freeness?
Is this mechanism better than the Permutation and Sidepayment mechanism?