Economics 395: Matching and Market Design, Winter 2023 Problem set 1
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
Economics 395: Matching and Market Design, Winter 2023
Problem set 1
1 Consider a public endowment problem with 6 individuals, {i1 ,i2 ,i3 ,i4 ,i5 ,i6 }, and 6 objects, {a1 ,a2 ,a3 ,a4 ,a5 ,a6 }. The individuals’ preferences over these objects are as follows:
Pi3
a3 a1 a2 a4 a5 a6 |
a3 a2 a1 a5 a4 a6 |
a1 a4 a3 a2 a6 a5 |
a2 a1 a5 a4 a3 a6 |
a2 a1 a6 a4 a5 a3 |
a1 a3 a2 a4 a6 a5 |
a . Give an example of a wasteful assignment (and explain why) .
b . Give an example of an inefficient assignment, di↵erent from that you found in a . (and explain why) .
c . Find the assignment produced by SD for the order i6 ,i5 , ...,i1 .
d . Show, without referring to SD, that the assignment you obtained in c . is efficient .
e . Suppose that i2 lies about her preference by reporting the following alter- native ranking:
i2
a5
a4
a3
a6
a1
a2
What’s the assignment produced by SD if i2 reports Pi\2 and every other in- dividual reports truthfully (i .e . , the above rankings)? Is i2 strictly better- o↵?
2. Five friends, Alice, Ben, Carly, Dan, and Emily all work together at the local favorite ice cream shop . As high school seniors, they each only have the abil- ity to work one of the five available time slots, labeled A through E . Their preferences for these time slots are illustrated in the table below .
PAlice PBen PCarly PDan PEmily
C D A C A
B B B A C
A E D E
D C D
E B
Remember that if an object is not listed in an individual’s preferences, they consider it unacceptable (in this example, this may possibly be due to schedul- ing constraints) .
a . Propose an assignment that fails individual rationality (IR) . Suppose we have the following assignment µ:
µAlice = C
µBen = D
µCarly = E
µDan = ;
µEmily = A
b . Is this assignment wasteful? Is this assignment IR? Is this assignment efficient? Explain each answer .
Suppose we made a slight adjustment and created a new assignment µ\ :
µ = C
µ = D
µ = A
µ = ;
µ = E
c . Do any of your answers to part b . change under this new assignment?
d . Find the assignment created by using the serial dictatorship (SD) algorithm with the following order: Dan, Alice, Emily, Ben, and Carly. Repeat part b . with this assignment .
3. Alice, Carly, Dan, and Emily all decide to attend State College (Ben instead leaves to attend City College) . As incoming freshman to State College, they apply for and enter the housing lottery over the schools five dorms: H1 , H2 , H3 , H4 , and H5 . For simplicity sake, assume each dorm can only house 1 student . As part of the application they rank their preferences over the dorms, shown below
PAlice PCarly PDan PEmily
H4 H2 H3 H5
H3 H1 H4 H4
H5 H3 H5 H1
H1 H4 H1 H2
H2 H5 H2 H3
a . Is it possible to construct a final assignment that does not satisfy IR? If “yes,” provide an example . If “no,” explain why not .
b . Notice that there are five potential dorms to be assigned but there are only four students to be assigned . Does this mean that any assignment is necessarily wasteful? Explain why or why not .
c . Intuitively, we can think of the following assignment as the “best” assign- ment:
µAlice = H4
µCarly = H2
µDan = H3
µEmily = H5
since every individual receives their top pick . Clearly this is an efficient assignment . Is this the only efficient assignment? If “yes,” explain why. If “no,” provide another efficient assignment .
4. Alice, Ben, Carly, and Dan do incredibly well in college and the four of them go to medical school and eventually become doctors . They coincidentally enter the labor market in the same area and form preferences over four potential hospitals W , X , Y , and Z, to work in . Management at each hospital also form priority orderings on the four doctors (including the possibility of determining one or more is unacceptable) .
PAlice PBen PCarly PDan
Y |
W |
W |
X |
X |
X |
Z |
Y |
Z |
Y |
X |
Z |
W |
Z |
Y |
W |
PW |
PX |
PY |
PZ |
Alice Dan Alice Ben
Dan Alice Carly Alice
Carly Carly Dan Carly
Ben Ben Ben Dan
a . Consider the following assignment: Alice is assigned to Y , Ben is assigned to Z, Carly is assigned to W, and Dan is assigned to X . Does this assignment lead to justified envy? Explain .
b . Consider a new assignment: Alice is assigned to W, Ben is assigned to Z , Carly is assigned to Y , and Dan is assigned to X . Does this assignment lead to justified envy? Explain .
c . Use the SD algorithm with the following order: Ben, Alice, Dan, Carly. What assignment does this result in? Does this assignment lead to justi- fied envy?
5. This problem revisits the notion of individual rationality in public endowment problems with priorities, when not all individuals are acceptable to all objects . Suppose that 5 public houses (objects) must be allocated among a group of 5 families (individuals), and imagine that the following tables describe rankings and priorities, where f denotes a generic family, and h a generic house:
Pf1 Pf2 Pf3 Pf4 Pf5
3 |
3 |
1 |
2 |
2 |
1 |
2 |
4 |
1 |
1 |
2 |
1 |
3 |
5 |
3 |
4 |
5 |
2 |
4 |
4 |
5 |
4 |
5 |
3 |
5 |
Ph1 |
Ph2 |
Ph3 |
Ph4 |
Ph5 |
f3 f1 |
f3 f2 f1 f5 |
f1 f4 f3 |
f2 f1 f5 f4 f3 |
f2 f1 f3 |
Remember that objects not listed on rankings, or individuals not listed on priorities, are considered unacceptable . One way to interpret these tables is to imagine that they capture, implicitly, some heterogeneity among the di↵erent houses (e .g . , number of rooms) and families (e .g . , number of members in a household .)
Consider the following alternative definition of individual rationality (IR), termed “IR for all:”
Definition 1. An assignment µ is IR for all if µi Pf ; for every f and µh ⇡h ; for every h,
where ⇡h denotes the priority of house h . In words, “IR for all” requires that no individual and object receives an assignment that is unacceptable .
a . Is “IR for all” stronger or weaker than IR? i .e . , must an assignment that is IR also be “IR for all” or the other way around? Why?
b . Does Serial Dictatorship (SD) satisfy “IR for all”? i .e . , does SD produce an assignment that is “IR for all” for every order of individuals? Why?
6. This problem invites you to think about an alternative notion of efficiency, termed “efficiency for all”, in public endowment problems with priorities . To define this new notion, let A denote the set of all individuals and objects (A stands for “agents .”) . Thus, a will denote either an individual or an object .
Definition 2. An assignment µ is efficient for all if there is no alternative assignment µ\ such that every agent strictly prefers µ\ to µ, or is indi↵erent, and there exists at least one agent that strictly prefers µ\ to µ .
Notice that this is the same definition of efficiency we discussed in class, with the exception that this definition is in terms of agents instead of individuals . Thus, this definition takes into account not only individuals’ rankings, but also objects’ priorities .
a . It turns out that “efficiency for all” is weaker than efficiency; i .e . , if an assignment is efficient, then it is efficient for all . Explain why, as precisely as you can, without using any particular example .
b . Is the converse true? i .e . , is it true that if an assignment is “efficient for all”, then it must be efficient? If “yes”, explain why. If “no”, provide an example .
7. This problem revisits the SD algorithm in public endowment problems with priorities . Consider a di↵erent version of SD in which objects are the ones picking individuals, given some order . Let’s term this new algorithm OSD (where O stands for ”object .”) Specifically, OSD fixes some order of objects, and objects choose in that order, among the individuals that were not picked before by some other object .
a . Does OSD satisfy “IR for all”? i .e . , does OSD produce an assignment that is “IR for all” for every order of objects? Why?
Consider now the same 5 public houses and 5 families from problem 5 . , and their same rankings and priorities .
b . Does OSD satisfy IR? i .e . , does OSD produce an assignment that is indi- vidually rational for every order of objects? Why? Compare your answer with the one you gave in 5 .b .
c . Is OSD efficient? i .e . , does OSD produce an efficient assignment for every order of objects?
2023-01-19