Exam 2 Flashcards
(79 cards)
Asymmetric Information
When one party of a transaction has more or better information than the other party
When sellers have more information than the buyers —> surplus-generating trades may not occur —> this is the market value
Could be transaction that would make both sides better, but asymmetric info prevents it from happening, as we will explain
Adverse selection
hidden fixed characteristics at the time of contracting
Fixed in the sense that they don’t change but you hide it (I.e. they don’t know about some characteristic that you have when you make the transaction (but they may learn it))
Ex: insurance - people with hidden characteristics buy it - someone who knows that they are sick could buy it - they know it but other side doesn’t
Ex: other side doesn’t know things like how bad a driver you are, how likely you are to get sick, or how smart/hard-working you are when you engage in a transaction with them
Moral hazard
hidden actions following contracting
Ex: insurance - after you buy insurance, you behave recklessly (I.e. buy car insurance and then drive more reckless)
Stuff that happens after you buy it,. such as questions about your investment strategy after getting funding or how hard you will work after receiving a loan
Lime/Lemon/Candy overview
2 groups of people - group 1 (citrus farmer) has a brown paper bag which has either a lemon or a lime (they don’t know); group 2 (juice producer) group has a mini-candy. Equal number of limes and lemons
group with the fruit (group 1) values a lime at 60 cents, a lemon at 30 cents, and a mini-candy at 50 cents
Group with the candy (group 2) values a lime at 70 cents, lemon at 40, and mini-candy at 50
So, for everyone a lime is better than a lemon; juice producer (group 2) values lime and lemon more than citrus farmer does
No-Info Case
Here, neither side has information about what is inside the bag - don’t know if lime or lemon
So, need to calculate expected value from a trade to see if worth it
To the farmer (group 1), EV (bag) = .50(.60)+.50(.30) = .45
To the juice producer (group 2), EV (bag) = .50(.70) + .50(.40) = .55
So, since the group without the bag values the bag more than the group with the bag, there can be trade for the mini candy (at 50 cents). Group 1 (sellers) will receive 50-45 = 5 cents and the buyers will receive 55-50=5 cents profit too.
If there were money, rather than just the mini-candies, could be trade between 45 and 55 cents, as this is the ZOPA - where everyone could benefit from a trade
Asymmetric Information Case
Now, same case, but the group with the bag (group 1) can see if it is a lemon or a lime
If price is a mini-candy (50 cents) - Since group 1 values limes more than candy, they will only want to trade the lemon. Group 2 won’t pay 50 cents for the lemon - need to lower the price. Market will unravel as no sale of limes. But, exchange for lemons between 30 and 40 cents could still occur.
Lime won’t sell for 60 cents, as group 1 can’t prove it is a lime, could be lying. Juice producer won’t pay 60 cents for this when expected value is 55 cents. Need a guarantee (if knew everything, then lime between 60 and 70 cents and lemon between 30 and 40) or no info - asymmetric information leads to a market failure in lack of trade when there should be trade here
Used Car argument overview
Adverse selection - consider a used car model, where the seller knows more about the car than the buyer.
Consider the buyer - ask why he is selling the car? Quality is worse than the price he’ll sell at. Buyer never wants to buy!
Akerlof Model Setup
Two groups of people
M is the stuff that people buy besides cars
xi is the quality of car i
Group 1:
Owns n cars; total income Y1
Total utility = U1 = M + from i=1 to n, the sum of xi
This means utility is what they can buy besides cars + the sum of the qualities of the cars they buy. One unit of utility for 1 unit of quality of the car
Group 2:
Owns 0 cars, total income Y2
Total utility = U2 = M + the sum from i=1 to n, 3/2* xi
This is saying the utility is what they can buy besides cars + 1.5 utility units for 1 unit of quality of car they buy. 50% more utility per quality than group 1
Group 1’s cars
n cars, think of n as large
Quality is uniformly distributed with values between 0 and 2
Price of M (Other uses of income besides cars) is 1; price of cars is p
Note: single price for all cars bc buyers can’t tell difference in quality
μ is the average quality of cars on market for sale
Group 1 D&S
If μ > p —> D1 = Y1/p
If the quality is greater than the price, then they spend all their money on cars
If μ<p> D1 = 0
If quality is less than price, then spend no money on cars, all on the other stuff
Average quality of cars on the market:
Start with recognizing that the quality of cars is uniformly distributed between 0 and 2. Imagine a number line with this distribution
Then realize that there is a price p between 0 and 2 - only car quality between 0 and p will come onto the market
So, the average quality will be the midway point between 0 and p = p/2 (note this 2 is different from the 2 being the upper range, but rather is showing it is the midway point between 0 and p)
So, average quality of cars on the market = μ = p/2
Supply of cars of group 1
They will supply cars with quality between 0 and p —> fraction of cars supplied will be the ones between 0 and p divided by the total number of cars (2). So, supply of cars = (p-0)/(2-0) = p/2.
This is the fraction of cars supplied - need to multiple this by n to get the total number of cars
This depends that the upper range of quality of cars on the market is 2
So, total supply of cars of group 1 —> If p≤2, then S1 = p*n/2 </p>
Group 2’s Demand and Supply
Demand: If 3μ/2 > p, then D2 = Y2/p If 3μ/2 < p, then D2 = 0 Either spend all their money on cars or none. The pivot point is 3μ/2 —> larger pivot point because value cars more Supply = 0 —> they don’t own any cars
Group 1+2 Demand
Get the aggregate demand - add up the demand functions for the two groups
If p Demand = (Y2+Y1)/p
Everyone wants to buy cars
If μ < p < 3μ/2 —> Demand = Y2/p
Here, only group 2 wants to buy cars, not group 1. Group 2 spend all their money on cars
If p > 3μ/2 —> Demand = 0
If price too high, nobody will demand any cars
Results of Akerlof Model
Remember that we know μ=p/2 from before, so we can plug this in to get the real demand
When we do this, only the third statement is true (because p isn’t less than μ or between μ and 1.5μ). Instead, p = 2μ, so the bottom equation is true
So, the total demand in the economy is 0
So, nobody will buy cars even though the group that owns cars should be selling cars as they generate less utility from a car than the group without car
So, no trade takes place despite buyers who value goods more than sellers
Average quality not worth justifying paying the price of p
—> When sellers have more information than buyers, surplus-generating trades may not occur. Only the bad types come onto the market, scare people off
Uzi, List, and Price Reading (Toward an Understanding of Why People Discriminate):
The disabled receives higher price quotes than the non-disabled receive
And the disparate treatment of the disabled is consistent with a model of statistical discrimination driven by search cost differences
Remedies for Market Failure Due to Adverse Selection:
Some market-based and some require regulation
Certification and Third Parties, Middle Men and Reputation, Asset Fire Sales (say fire at a warehouse, liquidate and sell for less - events that drive high quality goods on the market), government regulation requiring people to enter the market (health/car insurance), government regulation requiring more information to be made available, etc
Other Bad Market Outcomes from Adverse Selection:
Last class - surplus generating trades may not occur
Other costs of dishonesty -
Ex: time spend picking out stones from rice (efficiency lost)
Not just that not getting all the rice, but opportunity cost of losing time to do this
Statistical discrimination—> potential market response to adverse selection problem
Statistical Discrimination:
Statistical Discrimination: Idea that when lack of information, people may use information on group averages to make inferences about the group as a whole.
Group is defined by an easy-to-observe characteristic
Ex: May pay men and women differently —> don’t know how long each person will stay at the firm, but generally women stay less because have kid, so for all women, expect this and pay less
Ex: young men charged higher on car insurance bc more likely to crash
Ex: racial profiling at airports - could be more likely some races have guns
Profit maximizing but has the potential for negative consequences as can decrease the incentives to invest (why should women want to be in the workforce if paid less)
Distinct from taste-based discrimination
Taste-based is when you treat people differently because you like that group better. Preferences change more just because you don’t like them
Audit Study Methodology
Using trained actors to interact in real world markets
Advantage: Can try to hold constant all observable characteristics except the one key variable we are interested in
Compare outcomes in how the market treats them (prices offered, customers or applicants selected, etc)
Disadvantages:
Often not the real outcome of interest (may be that men get the job more often, but how about the middle step or vice versa)
Actors may know what the study is about (and this, unconsciously affects their behavior)
Ex: car salesman may notice half people in wheel chair and change how treated
Unobservable characteristics between groups not controlling for
Ex: send identical resumes and vary one characteristic like race signaled by the name. Examine call-back rates. Hold everything constant except the name
Audit Study on Discrimination Against the Disabled:
Reading from above where 6 disabled (in wheelchair) and 6 non-disabled people go to various body shops in Chicago to have their car bumpers fixed
Goal:
Look for evidence of discrimination against handicap people
Look for weather discrimination is taste-based or statistical
First experiment: “I’d like to fix my front bumper” then ask for a price quote
Testing for any discrimination - just looking to see if charged different amount
See that disabled are charged 593 while non-disabled charged 500 —> there is discrimination
Second experiment: Before asking for quote, say “I’m collecting a few quotes”
Testing for statistical discrimination based on search costs
Saying you are shopping around —> they assume person with wheelchair is looking around —> is difference in price based on these search costs.
Results: disabled group here for the least price - 461.4 and then non-disabled for 483.5
Overall conclusions:
Disabled given much higher quotes for repairing bumpers than non-disabled
Driven by statistical discrimination on basis of search costs rather than taste-based
This is found in other markets like mortgages (should always say shopping around so can get better price)
Moral Hazard and Real Estate Agents:
Standard contract w RE agent to sell home is pay agent 3% of price of the house.
Could be moral hazard as may not be worth the extra effort to get just a little more return
Ex: say you can sell house for 450-550k. If sell for 450k, then agent gets 15k and you 485k. If sell for 550k, agent gets 16.5 and you get 533k. So difference is just 1.5k for the agent but 48.5k for the owner —> lot of work for the agent to get just 1.5k more
We will look at what experts recommend/implement for others vs when they buy their own houses
This study only looks at the sale side
Empirical Evidence:
Levitt and Syverson look at 98k sales in suburban Chicago (3,300 houses owned and sold by agents for themselves)
Using regression to look at price of estate agents and compare with what sell their own houses for
Y = b*AGENT_OWNED + cX + e
b is how differently they behave when sell for themselves vs someone else
AGENT_OWNED is either 0 or 1 depending on who owns it
X is a vector of all characteristics of houses
c is a constant and e is the error
—> Agents’ own houses sell for 3.7% more and are on market for 9.5 extra days (10% longer)
We hold constant all other factors and see that b=3.7% when measuring cost and b=9.5 when measuring days.
Still is possible we didn’t control for everything or perhaps owners don’t listen to the agent
But, overall the evidence is there that moral hazard affects the selling behavior of real estate agents
But, does it matter that you are losing 3.7% and paying the agent the 3%? Is that $ important?
An alternative payout structure could be to pay a larger percent above certain amounts, but the pivot point is questionable and risky for the agent
Synthesis Class 2
Last session: model demonstrating a market failure from asymmetric information (adverse selection)
Private market for unemployment insurance - not good market for this because of adverse selection - only people likely to be fired would buy
Consider idea that statistical discrimination is a possible response to problems of adverse selection —> when individuals have more private information about their own quality or willingness to pay than the firm (wheelchair study)
Empirical evidence on moral hazard in RE markets
Peer Monitoring and Enforcement Reading:
“Group liability should improve repayment rates by providing incentives for peers to screen, monitor, and enforce each other’s loans,” but group liability may create excessive pressure and discourage good clients from borrowing, hurting growth and sustainability. No increase in default and larger groups - so group liability not working to help repayment rates - banks can do “just as well as peers at monitoring and enforcing loans and generating high repayment rates.”
Asymmetric Information in Creditor/Borrower Relationships:
Principal-agent relationships (p v a):
Bank vs bank loan borrower, firm issuing bonds vs bond investor, moneylender vs household, microfinance institution vs borrower
Adverse selection:
Hard for the bank to tell who is good vs bad type of borrower
Can charge higher interest rate for people who less likely to repay - but hard to know who they are
Moral hazard:
After borrow, do things that bank won’t want them to like investing in a more risky investment
Consider a borrower who can invest in 2 things and limited liability so the bank can’t punish the borrower if less than 0 return. In US, require collateral to prevent this, but not the case in small markets - bank can’t collateralize
Option 1: less risky —> 50% chance to get 100, 50% to get 0
If get the $100, then borrower gets $40 and bank gets $60 (interest).
If get the $0, then both get 0
Expected payoff to borrower is $20; to bank is $30
Bank likes this
Option 2: more risky —> 5% chance to get $1000, 95% chance to get $0
If get the $1000, borrower gets $940 and bank still gets $60
If get 0, both get 0
Expected payoff to borrower is 47, to thank is 3
Loan not equity contract, so all the upside goes to the borrower
So, the borrower clearly prefers the more risky while bank prefers less risky. So, huge issue here with moral hazard, as borrower will invest in risky things likely
Loan Screening and Monitoring:
Screening and monitoring to prevent adverse selection and moral hazard
Screening - seeing if good person beforehand (to mitigate adverse selection)
Monitoring - check in on them regularly (to prevent moral hazard)
Problem of lending to the poor - as high costs to gather information, small loan size, and loan officer not making a lot of money (about $1200) - responsible for 100 clients, so max you can spend on information collection is $12 per client
Issues with collateral - not available (poor property rights institutions) and not enforceable (poor legal institutions)