My Math Forum  

Go Back   My Math Forum > Science Forums > Economics

Economics Economics Forum - Financial Mathematics, Econometrics, Operations Research, Mathematical Finance, Computational Finance


Reply
 
LinkBack Thread Tools Display Modes
March 8th, 2010, 07:24 PM   #11
Newbie
 
Joined: Mar 2010

Posts: 11
Thanks: 0

Re: Fair Division

So funny, I just typed The Method of Sealed Bids into Google and wala, the first link was a great example of Fair Division (http://www.ctl.ua.edu/math103/FairDiv/s ... htm#SEALED BIDS EXAMPLE). Notice how Michael gets totally screwed. He gets $1,738 in cash, while Barbie gets $2,275 in items and $422 in cash. How is that fair? It sure would be nice if I could find something that spells out a scenario where everyone gets the same value.

In regards to your three examples, Iím not sure if I understand:

Example: The estate has a time machine (without power source), a time machine power source, and $1 billion. A player values the time machine at $1 million, the time machine power source at $1 million, and the combination of the two at $1 billion.

Sell the two items together. Your example is like selling a house and the land separately. Who would ever structure something like that?

Example: Two heirs are rivals, and gain positive utility from denying the other access to desired items (by taking them or causing them to go to another heir).

Whatever one bids on any item gets taken out of that personís share. So sure, you could deny someone access to an item but you end up compensating them for that decision.

Example: A painting is valued by A at $30,000, by B at $60,000, and by C at $90,000. The honest Vickrey valuation of the painting is $60,000. But if B and C know that A values the painting at $30,000 then B can dishonestly bid $30,000 for the painting and have C give him a side-payment between $10,000 and $30,000 (thus defrauding A).

Your right, if A chooses to tip his hand and let B and C know of his intended bid, collusion could occur. Tell A to keep his mouth shut.
fever is offline  
 
March 9th, 2010, 05:41 AM   #12
Global Moderator
 
CRGreathouse's Avatar
 
Joined: Nov 2006
From: UTC -5

Posts: 16,046
Thanks: 938

Math Focus: Number theory, computational mathematics, combinatorics, FOM, symbolic logic, TCS, algorithms
Re: Fair Division

Quote:
Originally Posted by fever
So funny, I just typed The Method of Sealed Bids into Google and wala, the first link was a great example of Fair Division (http://www.ctl.ua.edu/math103/FairDiv/s ... htm#SEALED BIDS EXAMPLE). Notice how Michael gets totally screwed. He gets $1,738 in cash, while Barbie gets $2,275 in items and $422 in cash. How is that fair?
Yes, how is it fair for Barbie? Michael thinks she got $6 worth of items and $422 in cash, for a total value of $428. By his valuation, he gets over four times more than her.

Quote:
Originally Posted by fever
Sell the two items together. Your example is like selling a house and the land separately. Who would ever structure something like that?
Yes, but that items have joint value is not always clear. For a more complicated example, look at the 2008 FCC spectrum auction.

Quote:
Originally Posted by fever
Example: Two heirs are rivals, and gain positive utility from denying the other access to desired items (by taking them or causing them to go to another heir).

Whatever one bids on any item gets taken out of that personís share. So sure, you could deny someone access to an item but you end up compensating them for that decision.
That means that it's possible. But consider an item which A values at only $100, but where A gains $1000 utility from denying the item to B. (Perhaps he's spiteful, or perhaps the item is noisy/garish/etc. and the two are neighbors...) If B values the item at $500 and C values the item at $300, the efficient solution is for A to pay C to place the top the top bid, and pay him $200 to $400 for it. But the mechanism doesn't make determining this joint valuation easy, nor does it determine a fair way for the surplus between A and C to be determined.

Quote:
Originally Posted by fever
Example: A painting is valued by A at $30,000, by B at $60,000, and by C at $90,000. The honest Vickrey valuation of the painting is $60,000. But if B and C know that A values the painting at $30,000 then B can dishonestly bid $30,000 for the painting and have C give him a side-payment between $10,000 and $30,000 (thus defrauding A).

Your right, if A chooses to tip his hand and let B and C know of his intended bid, collusion could occur. Tell A to keep his mouth shut.
It's frequently the case that heirs know each other well. If B or C can guess that A's bid will be $15,000 to $45,000, they can still collude by having B bid $15,000 (or $5,000). In fact, they don't even have to know that: if they trust each other, they could compare their bids and change the lower of their bids $0. If the higher bid wins, and the drop results in lowering the price of the item, they can agree to split the amount 50/50. (If they don't want to be suspected of collusion, they can reduce the bids, but by less than the full amount.)
CRGreathouse is offline  
March 9th, 2010, 06:59 AM   #13
Newbie
 
Joined: Mar 2010

Posts: 11
Thanks: 0

Re: Fair Division

Letís shelve the pro and con discussion of Vickeroy auctions and just say that all auctions have pros and cons. Moving on, I think I understand your response to my labeling of Michael as getting screwed. Correct me if Iím wrong, but youíre saying Michaelís realized value is actually greater than his perceived / utility value and Barbieís realized value is less. Perhaps itís my business background talking, but oneís perceived value should have nothing to do with an itemís market value. At the end of the day, each person should end up with an equal share of the same sized pie not different shares / different pies. Look at it this way; if all the items were sold for cash everyoneís share would be equal. Why, by bidding on items, do we suddenly give everyone different shares?
fever is offline  
March 9th, 2010, 07:37 AM   #14
Global Moderator
 
CRGreathouse's Avatar
 
Joined: Nov 2006
From: UTC -5

Posts: 16,046
Thanks: 938

Math Focus: Number theory, computational mathematics, combinatorics, FOM, symbolic logic, TCS, algorithms
Re: Fair Division

Quote:
Originally Posted by fever
Letís shelve the pro and con discussion of Vickeroy auctions and just say that all auctions have pros and cons.
I don't believe we've compared the Vickrey auction to any other auction type...

Quote:
Originally Posted by fever
Perhaps itís my business background talking, but oneís perceived value should have nothing to do with an itemís market value. At the end of the day, each person should end up with an equal share of the same sized pie not different shares / different pies.
If people value the items differently, then no matter what division system you use the people will rate the shares each person gets differently.

Quote:
Originally Posted by fever
Look at it this way; if all the items were sold for cash everyoneís share would be equal. Why, by bidding on items, do we suddenly give everyone different shares?
Because everyone values cash identically, but items differently. If each item is valued the same by all players, then everyone will get a share they agree is identical.


Example: 5 people, 7 items that everyone agrees are worth $1000, plus $3000 in cash; two people get two items and the other three get 1 item and $1000.
Example: As above, except that one person values the items at $500. He would be just as happy with $1000/$1500 as the above split (depending on which share he gets). To him, his fair share is (7 * $500 + $3000)/5 = $1300, or $800 plus an item, or $300 plus two items. By choosing the first of these, you can give him more than his fair share (by his standards) and give the others more than their fair share (by their standards). How should the surplus be shared? That's what the method has to decide.
(Same) example: 7 items are worth $500 on the open market, but four of the five place special value on the items; they would be willing to pay $1000 for them. If the items were all sold, each person would get $1300. Then the four could each try to buy the items back; depending on how this happens and whether they collude, they'll pay between $500 and $1000 for the items. (If I value an item at $500 but someone who values it at $1000 wants to buy it from me, I may very well sell for more than $500. If the items were auctioned, surely some of the items would go for more than $500.) Let's say that the average buyback price is $600. $400 was lost to outside sellers; this method is a way to keep that $400 internal and so give the players a Pareto improvement.
CRGreathouse is offline  
March 9th, 2010, 07:40 AM   #15
Global Moderator
 
CRGreathouse's Avatar
 
Joined: Nov 2006
From: UTC -5

Posts: 16,046
Thanks: 938

Math Focus: Number theory, computational mathematics, combinatorics, FOM, symbolic logic, TCS, algorithms
Re: Fair Division

Quote:
Originally Posted by fever
At the end of the day, each person should end up with an equal share of the same sized pie not different shares / different pies.
Let's do this differently. Take the example you posted and divide the goods by some method you prefer. Let's then compare the results.
CRGreathouse is offline  
March 9th, 2010, 08:25 AM   #16
Newbie
 
Joined: Mar 2010

Posts: 11
Thanks: 0

Re: Fair Division

Item Michael Barbie Jamie Who gets it
Watch $1 $200 $100 Barbie for $100
Camera $890 $500 $2500 Jamie for $890
Clock $5 $2000 $1000 Barbie for $1000
Sunglassess $0 $75 $55 Barbie for $55
Jacket $50 $50 $450 Jamie for $50
Jet Ski $1 $1000 $3000 Jamie for $1000

Total $0 $1,155 $1,940 $3,095
Fair Share $1032 $1032 $1032 $3,095
Settlement Receives $1032 Pays $123 Pays $908 All heirs receive the equivalent of $1,032

The above would be how I would split it, if conducting a Vickeroy Auction. I have no idea why Michael would bid $1 for a jet ski both in my system and the previous system.
fever is offline  
March 9th, 2010, 08:29 AM   #17
Newbie
 
Joined: Mar 2010

Posts: 11
Thanks: 0

Re: Fair Division

That didn't post very well. How about this one.

Item, Michael, Barbie, Jamie, Who gets it
Watch, $1, $200, $100, Barbie for $100
Camera, $890, $500, $2500, Jamie for $890
Clock , $5, $2000, $1000, Barbie for $1000
Sunglassess, $0, $75, $55, Barbie for $55
Jacket, $50, $50, $450, Jamie for $50
Jet Ski, $1, $1000, $3000, Jamie for $1000

Total, $0, $1,155, $1,940, $3,095
Fair Share, $1032, $1032, $1032, $3,095
Settlement
Michael Receives $1032
Barbie Pays $123
Jamie Pays $908
All heirs receive the equivalent of $1,032

The above would be how I would split it, if conducting a Vickeroy Auction. I have no idea why Michael would bid $1 for a jet ski both in my system and the previous system.
fever is offline  
March 9th, 2010, 09:57 AM   #18
Global Moderator
 
CRGreathouse's Avatar
 
Joined: Nov 2006
From: UTC -5

Posts: 16,046
Thanks: 938

Math Focus: Number theory, computational mathematics, combinatorics, FOM, symbolic logic, TCS, algorithms
Re: Fair Division

I see. I've used that method before myself, so I can't say I object to it. It's hard to say which is better; both methods give out goods to the player who wants them the most [assuming honest bids], so they're equivalent in a Coasian sense. It's not clear to me which method, if either, has claim to being the 'right' one -- nor even that such a method exists. Are you familiar with Arrow's General Possibility Theorem, or better yet the similar theorem of Gibbard & Satterthwaite?
CRGreathouse is offline  
March 9th, 2010, 06:16 PM   #19
Newbie
 
Joined: Mar 2010

Posts: 11
Thanks: 0

Re: Fair Division

I still canít rationalize Fair Division as being very fair. Choose whatever auction methodology you like, at the end of the bidding, the estate value should be based on the bids and split evenly amongst the bidders. Keep in mind, the focus really isnít on the winning bidder in either methodology; itís on how weíre calculating their respective shares/pie. I just did some research on Arrowí and Gibbard and I fail to see their application in this discussion but admittedly Iím not well versed in either theory.
fever is offline  
March 10th, 2010, 07:50 AM   #20
Global Moderator
 
CRGreathouse's Avatar
 
Joined: Nov 2006
From: UTC -5

Posts: 16,046
Thanks: 938

Math Focus: Number theory, computational mathematics, combinatorics, FOM, symbolic logic, TCS, algorithms
Re: Fair Division

Quote:
Originally Posted by fever
I still canít rationalize Fair Division as being very fair. Choose whatever auction methodology you like, at the end of the bidding, the estate value should be based on the bids and split evenly amongst the bidders.
I don't agree with that (even if I might be convinced to pick 'your' auction vs. 'their' auction). That division may result in the players being strictly less well-off than a different method; I think we should remain open to other possibilities.

Quote:
Originally Posted by fever
Keep in mind, the focus really isnít on the winning bidder in either methodology; itís on how weíre calculating their respective shares/pie.
Sensibly, since the auction methods don't change who wins from auction to auction.

Quote:
Originally Posted by fever
I just did some research on Arrowí and Gibbard and I fail to see their application in this discussion but admittedly Iím not well versed in either theory.
Arrow's theorem isn't that relevant, but it's better known than the other so I thought I'd suggest it.

Gibbard & Satterthwaite is actually extremely relevant, though the exact statement of the theorem likely does not apply. Gloss of the theorem: Take a 'reasonable' voting method. If there's no way to manipulate the system, then it is a dictatorship. It seems extremely likely that there are similar results that can be/have been proved relating to this division problem: that there are ways to manipulate the system by submitting dishonest/sophisticated bids. ("Bid" is dual to "vote", in this comparison.)

I've already given examples of how the players can manipulate various systems. So in light of that manipulation possibility, any conclusions you draw on the assumption that the bids are honest is questionable. Take, for example, your assertion that
Quote:
the estate value should be based on the bids and split evenly amongst the bidders
If the bids are honest, this may be a desirable principle (though I'm not convinced; see above). But when the votes are known to be dishonest, there's no reason to stick to that.

Say the estate has just one item. A values it at $90, B at $600, and C at $1200. First of all, it's not obvious that the Vickrey price of $600 is the right one -- any price from $600 to $1200 might lay claim to that. But what if B dishonestly bids $900? He knows that this will increase his share of the estate by $100, so it's in his interests. So just because we *want* to give everyone $200 doesn't mean that we should actually give everyone one-third of the second-highest bid, because that might not reflect reality.
CRGreathouse is offline  
Reply

  My Math Forum > Science Forums > Economics

Tags
division, fair



Search tags for this page
Click on a term to search for related topics.
Thread Tools
Display Modes


Similar Threads
Thread Thread Starter Forum Replies Last Post
Is it a fair game? mobel Advanced Statistics 2 December 18th, 2013 08:08 AM
fair division (lone-chooser) pkpeaches Applied Math 0 November 11th, 2012 10:48 AM
A Fair Coin Chikis Advanced Statistics 11 October 15th, 2012 10:02 AM
Fair die roll losm1 Advanced Statistics 4 April 5th, 2010 06:55 AM
Fair Division fever New Users 1 December 31st, 1969 04:00 PM





Copyright © 2019 My Math Forum. All rights reserved.