ECON 383 Practice Problems from Chapter 10 11, 12, 13, 14, 15 H. K. Chen (SFU) ECON 383 1 / 19 Chapter 10 — Ex.11 11. Parking spaces: {a, b, c}; people: {x, y, z}. The distance between a person j and a parking space i, di,j , is the number of blocks between i and j. j’s valuation for i is vi,j = 8 − di,j H. K. Chen (SFU) ECON 383 2 / 19 Chapter 10 — Ex.11 11. Parking spaces: {a, b, c}; people: {x, y, z}. The distance between a person j and a parking space i, di,j , is the number of blocks between i and j. j’s valuation for i is vi,j = 8 − di,j j x y z H. K. Chen (SFU) va,j ECON 383 vb,j vc,j 2 / 19 Chapter 10 — Ex.11 11. Parking spaces: {a, b, c}; people: {x, y, z}. The distance between a person j and a parking space i, di,j , is the number of blocks between i and j. j’s valuation for i is vi,j = 8 − di,j j x y z H. K. Chen (SFU) va,j 8−2 ECON 383 vb,j 8−3 vc,j 8−6 2 / 19 Chapter 10 — Ex.11 11. Parking spaces: {a, b, c}; people: {x, y, z}. The distance between a person j and a parking space i, di,j , is the number of blocks between i and j. j’s valuation for i is vi,j = 8 − di,j j x y z H. K. Chen (SFU) va,j 6 ECON 383 vb,j 5 vc,j 2 2 / 19 Chapter 10 — Ex.11 11. Parking spaces: {a, b, c}; people: {x, y, z}. The distance between a person j and a parking space i, di,j , is the number of blocks between i and j. j’s valuation for i is vi,j = 8 − di,j j x y z H. K. Chen (SFU) va,j 6 8−1 ECON 383 vb,j 5 8−2 vc,j 2 8−5 2 / 19 Chapter 10 — Ex.11 11. Parking spaces: {a, b, c}; people: {x, y, z}. The distance between a person j and a parking space i, di,j , is the number of blocks between i and j. j’s valuation for i is vi,j = 8 − di,j j x y z H. K. Chen (SFU) va,j 6 7 ECON 383 vb,j 5 6 vc,j 2 3 2 / 19 Chapter 10 — Ex.11 11. Parking spaces: {a, b, c}; people: {x, y, z}. The distance between a person j and a parking space i, di,j , is the number of blocks between i and j. j’s valuation for i is vi,j = 8 − di,j j x y z H. K. Chen (SFU) va,j 6 7 8−2 ECON 383 vb,j 5 6 8−1 vc,j 2 3 8−2 2 / 19 Chapter 10 — Ex.11 11. Parking spaces: {a, b, c}; people: {x, y, z}. The distance between a person j and a parking space i, di,j , is the number of blocks between i and j. j’s valuation for i is vi,j = 8 − di,j j x y z H. K. Chen (SFU) va,j 6 7 6 ECON 383 vb,j 5 6 7 vc,j 2 3 6 2 / 19 Chapter 10 — Ex.11(a) 11(a) Describe how you would set up this question as a matching problem. Who are the sellers/objects, who are the buyers? H. K. Chen (SFU) ECON 383 3 / 19 Chapter 10 — Ex.11(a) 11(a) Describe how you would set up this question as a matching problem. Who are the sellers/objects, who are the buyers? The sellers/objects are the parking spaces, {a, b, c} H. K. Chen (SFU) ECON 383 3 / 19 Chapter 10 — Ex.11(a) 11(a) Describe how you would set up this question as a matching problem. Who are the sellers/objects, who are the buyers? The sellers/objects are the parking spaces, {a, b, c} The buyers are the people living in apartments, {x, y, z} H. K. Chen (SFU) ECON 383 3 / 19 Chapter 10 — Ex.11(a) 11(a) Describe how you would set up this question as a matching problem. Who are the sellers/objects, who are the buyers? The sellers/objects are the parking spaces, {a, b, c} The buyers are the people living in apartments, {x, y, z} Buyers’ valuations are as calculated va,j j x 6 y 7 z 6 H. K. Chen (SFU) ECON 383 on the previous slide: vb,j vc,j 5 2 6 3 7 6 3 / 19 Chapter 10 — Ex.11(b) 11(b) Use the bipartite graph auction to determine the market-clearing prices H. K. Chen (SFU) ECON 383 4 / 19 Chapter 10 — Ex.11(b) 11(b) Use the bipartite graph auction to determine the market-clearing prices Round 1 pi Seller Buyer va,j , vb,j , vc,j Step 1. Set pi = 0 for all j H. K. Chen (SFU) ECON 383 0 a x 6, 5, 2 0 b y 7, 6, 3 0 c z 6, 7, 6 4 / 19 Chapter 10 — Ex.11(b) 11(b) Use the bipartite graph auction to determine the market-clearing prices Round 1 pi Seller Buyer va,j , vb,j , vc,j Step 1. Set pi = 0 for all j Step 2. Construct preferred seller graph H. K. Chen (SFU) ECON 383 0 a x 6, 5, 2 0 b y 7, 6, 3 0 c z 6, 7, 6 4 / 19 Chapter 10 — Ex.11(b) 11(b) Use the bipartite graph auction to determine the market-clearing prices Round 1 pi Seller Buyer va,j , vb,j , vc,j Step 1. Set pi = 0 for all j Step 2. Construct preferred seller graph Step 3. Note that {x, y} is constricted, and that {a} = N ({x, z}) Note that {x, y, z} is also a 0 a x 6, 5, 2 0 b y 7, 6, 3 0 c z 6, 7, 6 constriction, with {a, b} = N ({x, y, z}) H. K. Chen (SFU) ECON 383 4 / 19 Chapter 10 — Ex.11(b) 11(b) Use the bipartite graph auction to determine the market-clearing prices Step 4. Raise pa by 1 H. K. Chen (SFU) ECON 383 pi Seller Buyer va,j , vb,j , vc,j 1 a x 6, 5, 2 0 b y 7, 6, 3 0 c z 6, 7, 6 4 / 19 Chapter 10 — Ex.11(b) 11(b) Use the bipartite graph auction to determine the market-clearing prices Step 4. Raise pa by 1 Step 5. Normalize lowest price to 0 pi Seller Buyer va,j , vb,j , vc,j 1 a x 6, 5, 2 0 b y 7, 6, 3 0 c z 6, 7, 6 This step is redundant here. H. K. Chen (SFU) ECON 383 4 / 19 Chapter 10 — Ex.11(b) 11(b) Use the bipartite graph auction to determine the market-clearing prices Step 4. Raise pa by 1 Step 5. Normalize lowest price to 0 pi Seller Buyer va,j , vb,j , vc,j 1 a x 6, 5, 2 0 b y 7, 6, 3 0 c z 6, 7, 6 This step is redundant here. Round 2 Step 2. Construct preferred seller graph H. K. Chen (SFU) ECON 383 4 / 19 Chapter 10 — Ex.11(b) 11(b) Use the bipartite graph auction to determine the market-clearing prices Step 4. Raise pa by 1 Step 5. Normalize lowest price to 0 pi Seller Buyer va,j , vb,j , vc,j 1 a x 6, 5, 2 0 b y 7, 6, 3 0 c z 6, 7, 6 This step is redundant here. Round 2 Step 2. Construct preferred seller graph Step 3. {x, y, z} is constricted, with {a, b} = N ({x, y, z}) H. K. Chen (SFU) ECON 383 4 / 19 Chapter 10 — Ex.11(b) 11(b) Use the bipartite graph auction to determine the market-clearing prices Step 4. Raise pa and pb by 1 Step 5. Normalize lowest price to 0 pi Seller Buyer va,j , vb,j , vc,j 2 a x 6, 5, 2 1 b y 7, 6, 3 0 c z 6, 7, 6 This step is redundant here. H. K. Chen (SFU) ECON 383 4 / 19 Chapter 10 — Ex.11(b) 11(b) Use the bipartite graph auction to determine the market-clearing prices Step 4. Raise pa and pb by 1 Step 5. Normalize lowest price to 0 pi Seller Buyer va,j , vb,j , vc,j 2 a x 6, 5, 2 1 b y 7, 6, 3 0 c z 6, 7, 6 This step is redundant here. Round 3 Step 2. Construct preferred seller graph H. K. Chen (SFU) ECON 383 4 / 19 Chapter 10 — Ex.11(b) 11(b) Use the bipartite graph auction to determine the market-clearing prices Step 4. Raise pa and pb by 1 Step 5. Normalize lowest price to 0 pi Seller Buyer va,j , vb,j , vc,j 2 a x 6, 5, 2 1 b y 7, 6, 3 0 c z 6, 7, 6 This step is redundant here. Round 3 Step 2. Construct preferred seller graph A perfect matching is found. So (2, 1, 0) clears the market H. K. Chen (SFU) ECON 383 4 / 19 Chapter 10 — Ex.11(c) j x y z va,j 6 7 6 vb,j 5 6 7 vc,j 2 3 6 11(c) At a more informal level, how do the prices you determined for the parking spaces in (b) relate to these spaces’ intuitive “attractiveness” to the people in apartments x, y, and z? Explain. H. K. Chen (SFU) ECON 383 5 / 19 Chapter 10 — Ex.11(c) j x y z va,j 6 7 6 vb,j 5 6 7 vc,j 2 3 6 11(c) At a more informal level, how do the prices you determined for the parking spaces in (b) relate to these spaces’ intuitive “attractiveness” to the people in apartments x, y, and z? Explain. The market-clearing prices, (pa , pb , pc ) = (2, 1, 0), happens to be the number of buyer demanding the objects when they are free: a is demanded by both x and y when its price is zero b is demanded by z when its price is zero c is demanded by no one when its price is zero However, this phenomenon is not general. H. K. Chen (SFU) ECON 383 5 / 19 Chapter 10 — Ex.12 12. Two buyers’ valuations for objects a, b are as follows: Buyer j x y va,j 4 3 vb,j 1 2 Give three different sets of market-clearing prices for this market. Note that prices must be integers. H. K. Chen (SFU) ECON 383 6 / 19 Chapter 10 — Ex.12 12. Two buyers’ valuations for objects a, b are as follows: Buyer j x y va,j 4 3 vb,j 1 2 Give three different sets of market-clearing prices for this market. Note that prices must be integers. To clear the market, need to have a perfect matching, i.e. assigning different objects to different buyers H. K. Chen (SFU) ECON 383 6 / 19 Chapter 10 — Ex.12 12. Two buyers’ valuations for objects a, b are as follows: Buyer j x y va,j 4 3 vb,j 1 2 Give three different sets of market-clearing prices for this market. Note that prices must be integers. To clear the market, need to have a perfect matching, i.e. assigning different objects to different buyers Notice that both x and y prefer a to b; but x prefers a more “strongly” than y does H. K. Chen (SFU) ECON 383 6 / 19 Chapter 10 — Ex.12 12. Two buyers’ valuations for objects a, b are as follows: Buyer j x y va,j 4 3 vb,j 1 2 Give three different sets of market-clearing prices for this market. Note that prices must be integers. To clear the market, need to have a perfect matching, i.e. assigning different objects to different buyers Notice that both x and y prefer a to b; but x prefers a more “strongly” than y does x prefers a more “intensely” than y does: va,x > va,y H. K. Chen (SFU) ECON 383 6 / 19 Chapter 10 — Ex.12 12. Two buyers’ valuations for objects a, b are as follows: Buyer j x y va,j 4 3 vb,j 1 2 Give three different sets of market-clearing prices for this market. Note that prices must be integers. To clear the market, need to have a perfect matching, i.e. assigning different objects to different buyers Notice that both x and y prefer a to b; but x prefers a more “strongly” than y does x prefers a more “intensely” than y does: va,x > va,y x’s “relative valuation” for a is higher than y’s: va,x − vb,x > va,y − vb,y H. K. Chen (SFU) ECON 383 6 / 19 Chapter 10 — Ex.12 12. Two buyers’ valuations for objects a, b are as follows: Buyer j x y va,j 4 3 vb,j 1 2 Give three different sets of market-clearing prices for this market. Note that prices must be integers. To clear the market, need to have a perfect matching, i.e. assigning different objects to different buyers Notice that both x and y prefer a to b; but x prefers a more “strongly” than y does x prefers a more “intensely” than y does: va,x > va,y x’s “relative valuation” for a is higher than y’s: va,x − vb,x > va,y − vb,y Naturally then, the market-clearing prices should be such that a is assigned to x, and thus b is assigned to y Recall that market-clearing prices always lead to optimal matching H. K. Chen (SFU) ECON 383 6 / 19 Chapter 10 — Ex.12 Buyer j x y va,j 4 3 vb,j 1 2 To incentivize x and y to choose a and b respectively, the prices pa , pb must be such that H. K. Chen (SFU) 4 − pa ≥ 1 − pb (1) 3 − pa ≤ 2 − pb (2) pa ≤ 4 (3) pb ≤ 2 (4) ECON 383 7 / 19 Chapter 10 — Ex.12 Buyer j x y va,j 4 3 vb,j 1 2 To incentivize x and y to choose a and b respectively, the prices pa , pb must be such that 4 − pa ≥ 1 − pb (1) 3 − pa ≤ 2 − pb (2) pa ≤ 4 (3) pb ≤ 2 (4) (1) and (2) are equivalent to (by rearranging the terms): 1 + pb ≤ pa ≤ 3 + pb H. K. Chen (SFU) ECON 383 (5) 7 / 19 Chapter 10 — Ex.12 Therefore, any pa , pb that satisfy equations (3), (4), and (5) will clear the market: H. K. Chen (SFU) pa ≤ 4 (3) pb ≤ 2 (4) 1 + pb ≤ pa ≤ 3 + pb (5) ECON 383 8 / 19 Chapter 10 — Ex.12 Therefore, any pa , pb that satisfy equations (3), (4), and (5) will clear the market: pa ≤ 4 (3) pb ≤ 2 (4) 1 + pb ≤ pa ≤ 3 + pb (5) Take pb = 0, then any pa ∈ {1, 2, 3} would satisfy (3)–(5) So possible market-clearing prices (pa , pb ) are (1, 0), (2, 0), and (3, 0) H. K. Chen (SFU) ECON 383 8 / 19 Chapter 10 — Ex.12 Therefore, any pa , pb that satisfy equations (3), (4), and (5) will clear the market: pa ≤ 4 (3) pb ≤ 2 (4) 1 + pb ≤ pa ≤ 3 + pb (5) Take pb = 0, then any pa ∈ {1, 2, 3} would satisfy (3)–(5) So possible market-clearing prices (pa , pb ) are (1, 0), (2, 0), and (3, 0) Take pb = 1, then any pa ∈ {2, 3, 4} would satisfy (3)–(5) So possible market-clearing prices (pa , pb ) are (2, 1), (3, 1), and (4, 1) H. K. Chen (SFU) ECON 383 8 / 19 Chapter 10 — Ex.12 Therefore, any pa , pb that satisfy equations (3), (4), and (5) will clear the market: pa ≤ 4 (3) pb ≤ 2 (4) 1 + pb ≤ pa ≤ 3 + pb (5) Take pb = 0, then any pa ∈ {1, 2, 3} would satisfy (3)–(5) So possible market-clearing prices (pa , pb ) are (1, 0), (2, 0), and (3, 0) Take pb = 1, then any pa ∈ {2, 3, 4} would satisfy (3)–(5) So possible market-clearing prices (pa , pb ) are (2, 1), (3, 1), and (4, 1) Take pb = 2, then any pa ∈ {3, 4} would satisfy (3)–(5) So possible market-clearing prices (pa , pb ) are (3, 2) and (4, 2) H. K. Chen (SFU) ECON 383 8 / 19 Chapter 10 — Ex.13(a) 13. You have two units of a good, and there are four potential buyers who value each unit at vi (each buyer wants at most one unit). (a) Describe how you would sell these two units of goods using an ascending-bid auction. H. K. Chen (SFU) ECON 383 9 / 19 Chapter 10 — Ex.13(a) 13. You have two units of a good, and there are four potential buyers who value each unit at vi (each buyer wants at most one unit). (a) Describe how you would sell these two units of goods using an ascending-bid auction. Call the two units of the good g1 , g2 H. K. Chen (SFU) ECON 383 9 / 19 Chapter 10 — Ex.13(a) 13. You have two units of a good, and there are four potential buyers who value each unit at vi (each buyer wants at most one unit). (a) Describe how you would sell these two units of goods using an ascending-bid auction. Call the two units of the good g1 , g2 Label the four buyers as 1, 2, 3, 4, and assume, without loss of generality, that v1 = 1, v2 = 2, v3 = 3, v4 = 4. H. K. Chen (SFU) ECON 383 9 / 19 Chapter 10 — Ex.13(a) 13. You have two units of a good, and there are four potential buyers who value each unit at vi (each buyer wants at most one unit). (a) Describe how you would sell these two units of goods using an ascending-bid auction. Call the two units of the good g1 , g2 Label the four buyers as 1, 2, 3, 4, and assume, without loss of generality, that v1 = 1, v2 = 2, v3 = 3, v4 = 4. As there are more buyers than objects, we can model this as a matching problem by adding two dummy objects, d1 , d2 , of which each buyer has a zero valuation H. K. Chen (SFU) ECON 383 9 / 19 Chapter 10 — Ex.13(a) The matching problem can be represented by a graph H. K. Chen (SFU) ECON 383 p Object Buyer valuations g1 1 1, 1, 0, 0 g2 2 2, 2, 0, 0 d1 3 3, 3, 0, 0 d2 4 4, 4, 0, 0 10 / 19 Chapter 10 — Ex.13(a) The matching problem can be represented by a graph p Object Buyer valuations Run the bipartite graph auction 0 g1 1 1, 1, 0, 0 0 g2 2 2, 2, 0, 0 0 d1 3 3, 3, 0, 0 0 d2 4 4, 4, 0, 0 H. K. Chen (SFU) ECON 383 10 / 19 Chapter 10 — Ex.13(a) The matching problem can be represented by a graph p Object Buyer valuations Run the bipartite graph auction Round 1 fails to produce a perfect matching 0 g1 1 1, 1, 0, 0 0 g2 2 2, 2, 0, 0 There are 5 constricted sets 0 d1 3 3, 3, 0, 0 0 d2 4 4, 4, 0, 0 H. K. Chen (SFU) ECON 383 10 / 19 Chapter 10 — Ex.13(a) The matching problem can be represented by a graph p Object Buyer valuations Run the bipartite graph auction Round 1 fails to produce a perfect matching 1 g1 1 1, 1, 0, 0 1 g2 2 2, 2, 0, 0 There are 5 constricted sets 0 d1 3 3, 3, 0, 0 0 d2 4 4, 4, 0, 0 Round 2 also fails to produce a perfect matching There is 1 constricted set H. K. Chen (SFU) ECON 383 10 / 19 Chapter 10 — Ex.13(a) The matching problem can be represented by a graph p Object Buyer valuations Run the bipartite graph auction Round 1 fails to produce a perfect matching 2 g1 1 1, 1, 0, 0 2 g2 2 2, 2, 0, 0 There are 5 constricted sets 0 d1 3 3, 3, 0, 0 0 d2 4 4, 4, 0, 0 Round 2 also fails to produce a perfect matching There is 1 constricted set Round 3 has a perfect matching H. K. Chen (SFU) ECON 383 10 / 19 Chapter 10 — Ex.13(a) The matching problem can be represented by a graph p Object Buyer valuations Run the bipartite graph auction Round 1 fails to produce a perfect matching 2 g1 1 1, 1, 0, 0 2 g2 2 2, 2, 0, 0 There are 5 constricted sets 0 d1 3 3, 3, 0, 0 0 d2 4 4, 4, 0, 0 Round 2 also fails to produce a perfect matching There is 1 constricted set Round 3 has a perfect matching H. K. Chen (SFU) ECON 383 10 / 19 Chapter 10 — Ex.13(a) The matching problem can be represented by a graph p Object Buyer valuations Run the bipartite graph auction Round 1 fails to produce a perfect matching 2 g1 1 1, 1, 0, 0 2 g2 2 2, 2, 0, 0 There are 5 constricted sets 0 d1 3 3, 3, 0, 0 0 d2 4 4, 4, 0, 0 Round 2 also fails to produce a perfect matching There is 1 constricted set Round 3 has a perfect matching Therefore the market-clearing price for each unit of the object is 2 H. K. Chen (SFU) ECON 383 10 / 19 Chapter 10 — Ex.13(b) 13(b) In the case of single-item auction, the bipartite graph procedure yielded the simple rule from the ascending-bid (English) auction: Rule for single-item auction Sell to the highest bidder at the second-highest price. In simple terms, what should be the rule for the current case of two identical items? H. K. Chen (SFU) ECON 383 11 / 19 Chapter 10 — Ex.13(b) 13(b) In the case of single-item auction, the bipartite graph procedure yielded the simple rule from the ascending-bid (English) auction: Rule for single-item auction Sell to the highest bidder at the second-highest price. In simple terms, what should be the rule for the current case of two identical items? Rule for two-identical-item auction Sell to the two highest bidders at the third-highest price. H. K. Chen (SFU) ECON 383 11 / 19 Chapter 10 — Ex.13(b) 13(b) In the case of single-item auction, the bipartite graph procedure yielded the simple rule from the ascending-bid (English) auction: Rule for single-item auction Sell to the highest bidder at the second-highest price. In simple terms, what should be the rule for the current case of two identical items? Rule for two-identical-item auction Sell to the two highest bidders at the third-highest price. Can you conjecture the rule for general n-item auction? H. K. Chen (SFU) ECON 383 11 / 19 Chapter 10 — Ex.13(b) 13(b) In the case of single-item auction, the bipartite graph procedure yielded the simple rule from the ascending-bid (English) auction: Rule for single-item auction Sell to the highest bidder at the second-highest price. In simple terms, what should be the rule for the current case of two identical items? Rule for two-identical-item auction Sell to the two highest bidders at the third-highest price. Can you conjecture the rule for general n-item auction? Rule for n-identical-item auction Sell to the n highest bidders at the (n + 1)th -highest price. H. K. Chen (SFU) ECON 383 11 / 19 Chapter 10 — Ex.14 14. Recall that A matching M is social welfare maximizing if it maximizes the sum of buyers’ valuations for what they get, over all possible perfect matchings; i.e. M is the solution to max ∑ vm(i),i m H. K. Chen (SFU) i ECON 383 12 / 19 Chapter 10 — Ex.14 14. Recall that A matching M is social welfare maximizing if it maximizes the sum of buyers’ valuations for what they get, over all possible perfect matchings; i.e. M is the solution to max ∑ vm(i),i m i However, the sum of buyers’ valuations need not be the only criterion that we use to judge the desirability of a matching. Sometimes we may want to have a matching that ensures that no buyer gets a valuation that is too small. H. K. Chen (SFU) ECON 383 12 / 19 Chapter 10 — Ex.14 Define the baseline of a matching m to be the minimum valuation that any buyer has for the item they get in M; i.e. min{vm(i),i |i = 1, . . . , n}. A matching M is baseline maximizing if M is the solution to n o max min{vm(i),i |i = 1, . . . , n} m H. K. Chen (SFU) ECON 383 13 / 19 Chapter 10 — Ex.14 Define the baseline of a matching m to be the minimum valuation that any buyer has for the item they get in M; i.e. min{vm(i),i |i = 1, . . . , n}. A matching M is baseline maximizing if M is the solution to n o max min{vm(i),i |i = 1, . . . , n} m This notion of baseline is motivated by egalitarian considerations: no one should be left too badly off. H. K. Chen (SFU) ECON 383 13 / 19 Chapter 10 — Ex.14 Buyer i x y z vb,j 7 9 10 va,j 9 5 11 vc,j 4 7 8 For example, consider a matching M = {a-x, b-y, c-z} The baseline of M is 8, because 8 = min{9, 9, 8} The social welfare of M is 9 + 9 + 8 = 26 H. K. Chen (SFU) ECON 383 14 / 19 Chapter 10 — Ex.14 Buyer i x y z vb,j 7 9 10 va,j 9 5 11 vc,j 4 7 8 For example, consider a matching M = {a-x, b-y, c-z} The baseline of M is 8, because 8 = min{9, 9, 8} The social welfare of M is 9 + 9 + 8 = 26 Consider another matching M0 = {b-x, c-y, a-z} The baseline of M0 is 7, because 7 = min{7, 7, 11} The social welfare of M0 is 7 + 7 + 11 = 25 H. K. Chen (SFU) ECON 383 14 / 19 Chapter 10 — Ex.14 Buyer i x y z vb,j 7 9 10 va,j 9 5 11 vc,j 4 7 8 For example, consider a matching M = {a-x, b-y, c-z} The baseline of M is 8, because 8 = min{9, 9, 8} The social welfare of M is 9 + 9 + 8 = 26 Consider another matching M0 = {b-x, c-y, a-z} The baseline of M0 is 7, because 7 = min{7, 7, 11} The social welfare of M0 is 7 + 7 + 11 = 25 M is also the baseline maximizing matching H. K. Chen (SFU) ECON 383 14 / 19 Chapter 10 — Ex.14(a) 14(a) Give an example of equal-sized sets of sellers and buyers, with valuations on the buyers, so that there is no perfect matching that is both social welfare maximizing and baseline maximizing. (In other words, social-welfare maximization and baseline maximization should only occur with different matchings.) H. K. Chen (SFU) ECON 383 15 / 19 Chapter 10 — Ex.14(a) 14(a) Give an example of equal-sized sets of sellers and buyers, with valuations on the buyers, so that there is no perfect matching that is both social welfare maximizing and baseline maximizing. (In other words, social-welfare maximization and baseline maximization should only occur with different matchings.) Consider a two-seller (a, b) and two-buyer (x, y) example. H. K. Chen (SFU) ECON 383 15 / 19 Chapter 10 — Ex.14(a) 14(a) Give an example of equal-sized sets of sellers and buyers, with valuations on the buyers, so that there is no perfect matching that is both social welfare maximizing and baseline maximizing. (In other words, social-welfare maximization and baseline maximization should only occur with different matchings.) Consider a two-seller (a, b) and two-buyer (x, y) example. There are only two possible perfect matchings H. K. Chen (SFU) ECON 383 15 / 19 Chapter 10 — Ex.14(a) 14(a) Give an example of equal-sized sets of sellers and buyers, with valuations on the buyers, so that there is no perfect matching that is both social welfare maximizing and baseline maximizing. (In other words, social-welfare maximization and baseline maximization should only occur with different matchings.) Consider a two-seller (a, b) and two-buyer (x, y) example. There are only two possible perfect matchings M = {a-x, b-y} H. K. Chen (SFU) ECON 383 15 / 19 Chapter 10 — Ex.14(a) 14(a) Give an example of equal-sized sets of sellers and buyers, with valuations on the buyers, so that there is no perfect matching that is both social welfare maximizing and baseline maximizing. (In other words, social-welfare maximization and baseline maximization should only occur with different matchings.) Consider a two-seller (a, b) and two-buyer (x, y) example. There are only two possible perfect matchings M = {a-x, b-y} M0 = {b-x, a-y} H. K. Chen (SFU) ECON 383 15 / 19 Chapter 10 — Ex.14(a) 14(a) Give an example of equal-sized sets of sellers and buyers, with valuations on the buyers, so that there is no perfect matching that is both social welfare maximizing and baseline maximizing. (In other words, social-welfare maximization and baseline maximization should only occur with different matchings.) Consider a two-seller (a, b) and two-buyer (x, y) example. There are only two possible perfect matchings M = {a-x, b-y} M0 = {b-x, a-y} Our objective is to make M (only) social welfare maximizing, and make M0 (only) baseline maximizing H. K. Chen (SFU) ECON 383 15 / 19 Chapter 10 — Ex.14(a) 14(a) Give an example of equal-sized sets of sellers and buyers, with valuations on the buyers, so that there is no perfect matching that is both social welfare maximizing and baseline maximizing. (In other words, social-welfare maximization and baseline maximization should only occur with different matchings.) Consider a two-seller (a, b) and two-buyer (x, y) example. There are only two possible perfect matchings M = {a-x, b-y} M0 = {b-x, a-y} Our objective is to make M (only) social welfare maximizing, and make M0 (only) baseline maximizing Note that it’s the same if you make M baseline maximizing and M0 social welfare maximizing H. K. Chen (SFU) ECON 383 15 / 19 Chapter 10 — Ex.14(a) M is social welfare maximizing implies va,x + vb,y > vb,x + va,y H. K. Chen (SFU) ECON 383 (1) 16 / 19 Chapter 10 — Ex.14(a) M is social welfare maximizing implies va,x + vb,y > vb,x + va,y (1) M0 is baseline maximizing implies min{va,x , vb,y } < min{vb,x , va,y } H. K. Chen (SFU) ECON 383 (2) 16 / 19 Chapter 10 — Ex.14(a) M is social welfare maximizing implies va,x + vb,y > vb,x + va,y (1) M0 is baseline maximizing implies min{va,x , vb,y } < min{vb,x , va,y } (2) Suppose min{va,x , vb,y } = va,x = 0 and min{vb,x , va,y } = vb,x = 1, so that condition (2) is satisfied. H. K. Chen (SFU) ECON 383 16 / 19 Chapter 10 — Ex.14(a) M is social welfare maximizing implies va,x + vb,y > vb,x + va,y (1) M0 is baseline maximizing implies min{va,x , vb,y } < min{vb,x , va,y } (2) Suppose min{va,x , vb,y } = va,x = 0 and min{vb,x , va,y } = vb,x = 1, so that condition (2) is satisfied. Condition (1) becomes 0 + vb,y > 1 + va,y H. K. Chen (SFU) ECON 383 (1’) 16 / 19 Chapter 10 — Ex.14(a) M is social welfare maximizing implies va,x + vb,y > vb,x + va,y (1) M0 is baseline maximizing implies min{va,x , vb,y } < min{vb,x , va,y } (2) Suppose min{va,x , vb,y } = va,x = 0 and min{vb,x , va,y } = vb,x = 1, so that condition (2) is satisfied. Condition (1) becomes 0 + vb,y > 1 + va,y (1’) Need to choose vb,y > 0 and va,y > 1 to satisfy (1’) H. K. Chen (SFU) ECON 383 16 / 19 Chapter 10 — Ex.14(a) M is social welfare maximizing implies va,x + vb,y > vb,x + va,y (1) M0 is baseline maximizing implies min{va,x , vb,y } < min{vb,x , va,y } (2) Suppose min{va,x , vb,y } = va,x = 0 and min{vb,x , va,y } = vb,x = 1, so that condition (2) is satisfied. Condition (1) becomes 0 + vb,y > 1 + va,y (1’) Need to choose vb,y > 0 and va,y > 1 to satisfy (1’) Pick va,y = 2 so that min{vb,x , va,y } = vb,x = 1 is satisfied H. K. Chen (SFU) ECON 383 16 / 19 Chapter 10 — Ex.14(a) M is social welfare maximizing implies va,x + vb,y > vb,x + va,y (1) M0 is baseline maximizing implies min{va,x , vb,y } < min{vb,x , va,y } (2) Suppose min{va,x , vb,y } = va,x = 0 and min{vb,x , va,y } = vb,x = 1, so that condition (2) is satisfied. Condition (1) becomes 0 + vb,y > 1 + va,y (1’) Need to choose vb,y > 0 and va,y > 1 to satisfy (1’) Pick va,y = 2 so that min{vb,x , va,y } = vb,x = 1 is satisfied Pick vb,y = 4 so that (1’) holds H. K. Chen (SFU) ECON 383 16 / 19 Chapter 10 — Ex.14(a) We can verify that (va,x , vb,x , va,y , vb,y ) = (0, 1, 2, 4) lead to distinct matchings for the two maximizations H. K. Chen (SFU) ECON 383 17 / 19 Chapter 10 — Ex.14(a) We can verify that (va,x , vb,x , va,y , vb,y ) = (0, 1, 2, 4) lead to distinct matchings for the two maximizations Under M = {a-x, b-y}, H. K. Chen (SFU) ECON 383 17 / 19 Chapter 10 — Ex.14(a) We can verify that (va,x , vb,x , va,y , vb,y ) = (0, 1, 2, 4) lead to distinct matchings for the two maximizations Under M = {a-x, b-y}, social welfare is 0 + 4 = 4 H. K. Chen (SFU) ECON 383 17 / 19 Chapter 10 — Ex.14(a) We can verify that (va,x , vb,x , va,y , vb,y ) = (0, 1, 2, 4) lead to distinct matchings for the two maximizations Under M = {a-x, b-y}, social welfare is 0 + 4 = 4 baseline is min{0, 4} = 0 H. K. Chen (SFU) ECON 383 17 / 19 Chapter 10 — Ex.14(a) We can verify that (va,x , vb,x , va,y , vb,y ) = (0, 1, 2, 4) lead to distinct matchings for the two maximizations Under M = {a-x, b-y}, social welfare is 0 + 4 = 4 baseline is min{0, 4} = 0 Under M0 = {b-x, a-y}, H. K. Chen (SFU) ECON 383 17 / 19 Chapter 10 — Ex.14(a) We can verify that (va,x , vb,x , va,y , vb,y ) = (0, 1, 2, 4) lead to distinct matchings for the two maximizations Under M = {a-x, b-y}, social welfare is 0 + 4 = 4 baseline is min{0, 4} = 0 Under M0 = {b-x, a-y}, social welfare is 1 + 2 = 3 H. K. Chen (SFU) ECON 383 17 / 19 Chapter 10 — Ex.14(a) We can verify that (va,x , vb,x , va,y , vb,y ) = (0, 1, 2, 4) lead to distinct matchings for the two maximizations Under M = {a-x, b-y}, social welfare is 0 + 4 = 4 baseline is min{0, 4} = 0 Under M0 = {b-x, a-y}, social welfare is 1 + 2 = 3 baseline is min{1, 2} = 1 H. K. Chen (SFU) ECON 383 17 / 19 Chapter 10 — Ex.14(a) We can verify that (va,x , vb,x , va,y , vb,y ) = (0, 1, 2, 4) lead to distinct matchings for the two maximizations Under M = {a-x, b-y}, social welfare is 0 + 4 = 4 baseline is min{0, 4} = 0 Under M0 = {b-x, a-y}, social welfare is 1 + 2 = 3 baseline is min{1, 2} = 1 Thus we have achieved the objective: M is social welfare maximizing while M0 is baseline maximizing H. K. Chen (SFU) ECON 383 17 / 19 Chapter 10 — Ex.14(b) 14(b) Give a yes/no answer to the following question. Explain your answer. For any equal-sized sets of sellers and buyers, with valuations on the buyers, is there always a set of market-clearing prices so that the resulting preferred-seller graph contains a baseline-maximizing perfect matching? H. K. Chen (SFU) ECON 383 18 / 19 Chapter 10 — Ex.14(b) 14(b) Give a yes/no answer to the following question. Explain your answer. For any equal-sized sets of sellers and buyers, with valuations on the buyers, is there always a set of market-clearing prices so that the resulting preferred-seller graph contains a baseline-maximizing perfect matching? No. The example in part (a) is one where no market-clearing prices can produce a perfect matching that is baseline maximizing. H. K. Chen (SFU) ECON 383 18 / 19 Chapter 10 — Ex.14(b) 14(b) Give a yes/no answer to the following question. Explain your answer. For any equal-sized sets of sellers and buyers, with valuations on the buyers, is there always a set of market-clearing prices so that the resulting preferred-seller graph contains a baseline-maximizing perfect matching? No. The example in part (a) is one where no market-clearing prices can produce a perfect matching that is baseline maximizing. Recall that (va,x , vb,x , va,y , vb,y ) = (0, 1, 2, 4) and M0 = {b-x, a-y} M = {a-x, b-y} H. K. Chen (SFU) ECON 383 18 / 19 Chapter 10 — Ex.14(b) 14(b) Give a yes/no answer to the following question. Explain your answer. For any equal-sized sets of sellers and buyers, with valuations on the buyers, is there always a set of market-clearing prices so that the resulting preferred-seller graph contains a baseline-maximizing perfect matching? No. The example in part (a) is one where no market-clearing prices can produce a perfect matching that is baseline maximizing. Recall that (va,x , vb,x , va,y , vb,y ) = (0, 1, 2, 4) and M0 = {b-x, a-y} M = {a-x, b-y} Any set of market-clearing prices must lead to a social welfare maximizing matching, which is M in this example; while M0 is the only baseline maximizing matching. H. K. Chen (SFU) ECON 383 18 / 19 Chapter 10 — Ex.15 15. Consider a bipartite graph auction with equal number of sellers and buyers. Suppose a particular seller i is every buyer’s favorite, in the sense that vi,j ≥ vk,j for every buyer j and every other seller k. Given a set of market-clearing prices, must it be the case that the price charged by seller i is at least as high as the price charged by any other seller (i.e. pi ≥ pk for any other seller k)? H. K. Chen (SFU) ECON 383 19 / 19 Chapter 10 — Ex.15 15. Consider a bipartite graph auction with equal number of sellers and buyers. Suppose a particular seller i is every buyer’s favorite, in the sense that vi,j ≥ vk,j for every buyer j and every other seller k. Given a set of market-clearing prices, must it be the case that the price charged by seller i is at least as high as the price charged by any other seller (i.e. pi ≥ pk for any other seller k)? Suppose (for contradiction) there exists a seller a whose price pa > pi H. K. Chen (SFU) ECON 383 19 / 19 Chapter 10 — Ex.15 15. Consider a bipartite graph auction with equal number of sellers and buyers. Suppose a particular seller i is every buyer’s favorite, in the sense that vi,j ≥ vk,j for every buyer j and every other seller k. Given a set of market-clearing prices, must it be the case that the price charged by seller i is at least as high as the price charged by any other seller (i.e. pi ≥ pk for any other seller k)? Suppose (for contradiction) there exists a seller a whose price pa > pi Since prices clear the market, there is a buyer x who gets object a H. K. Chen (SFU) ECON 383 19 / 19 Chapter 10 — Ex.15 15. Consider a bipartite graph auction with equal number of sellers and buyers. Suppose a particular seller i is every buyer’s favorite, in the sense that vi,j ≥ vk,j for every buyer j and every other seller k. Given a set of market-clearing prices, must it be the case that the price charged by seller i is at least as high as the price charged by any other seller (i.e. pi ≥ pk for any other seller k)? Suppose (for contradiction) there exists a seller a whose price pa > pi Since prices clear the market, there is a buyer x who gets object a Market-clearing also dictates that x weakly prefers a to j: va,x − pa ≥ vi,x − pi H. K. Chen (SFU) ⇒ (va,x − vi,x ) + (pi − pa ) ≥ 0 ECON 383 19 / 19 Chapter 10 — Ex.15 15. Consider a bipartite graph auction with equal number of sellers and buyers. Suppose a particular seller i is every buyer’s favorite, in the sense that vi,j ≥ vk,j for every buyer j and every other seller k. Given a set of market-clearing prices, must it be the case that the price charged by seller i is at least as high as the price charged by any other seller (i.e. pi ≥ pk for any other seller k)? Suppose (for contradiction) there exists a seller a whose price pa > pi Since prices clear the market, there is a buyer x who gets object a Market-clearing also dictates that x weakly prefers a to j: va,x − pa ≥ vi,x − pi ⇒ (va,x − vi,x ) + (pi − pa ) ≥ 0 But va,x ≤ vi,x and pi < pa , so a contradiction! H. K. Chen (SFU) ECON 383 19 / 19 Chapter 10 — Ex.15 15. Consider a bipartite graph auction with equal number of sellers and buyers. Suppose a particular seller i is every buyer’s favorite, in the sense that vi,j ≥ vk,j for every buyer j and every other seller k. Given a set of market-clearing prices, must it be the case that the price charged by seller i is at least as high as the price charged by any other seller (i.e. pi ≥ pk for any other seller k)? Suppose (for contradiction) there exists a seller a whose price pa > pi Since prices clear the market, there is a buyer x who gets object a Market-clearing also dictates that x weakly prefers a to j: va,x − pa ≥ vi,x − pi ⇒ (va,x − vi,x ) + (pi − pa ) ≥ 0 But va,x ≤ vi,x and pi < pa , so a contradiction! Therefore it’s impossible to find a seller a with pa > pi . So the statement in the question is true. H. K. Chen (SFU) ECON 383 19 / 19