The Devil You Know vs The Market For Lemons

I told my friend Mike I was thinking of trading my car in for one that used less gas. Nowadays I would talk about the Market for Lemons as a reason against doing what I was thinking of doing. But in 1974 I didn't know about Akerlof's famous paper, and neither did Mike, who was majoring in music. "Better to stick with the devil you know" he said.

"The Devil you know" and "the Market for Lemons" are similar: both use asymmetric information (the seller knows more about the car than the prospective buyer) to explain a low volume of trade. But they are not the same. A thought-experiment can explain the difference.

Suppose the government made a surprise announcement: everybody who owns a used car must sell it next week to the highest bidder. If you still want a car you must sell yours and buy someone else's.

Making sales compulsory solves the Market for Lemons problem. The sample of cars offered for sale represents the whole population of cars; those who know their own car is a cream puff can't hold it back from the market. If the Market for Lemons was the only problem in the used car market, then the compulsory universal sale would improve efficiency in the allocation of cars. People who found themselves needing to drive more miles could sell their gas guzzlers to the people who found themselves driving fewer miles.

But universal compulsory sales of cars does not solve the Devil You Know problem.

First there's simple risk aversion. Suppose everyone knows that half the cars need a $1,000 repair and the other half don't, but only the seller knows if his own car needs that repair. Sellers of lemons have a 50% chance of winning $1,000, which is worth less than a certain $500 if they are risk-averse. Sellers of cream puffs know they have a 50% chance of losing $1,000, which is worth more than a certain $500 loss if they are risk-averse. (Remember that everyone buys a car for the same price they sell their car, so price is a wash, if this is the only difference.) The value of the expected gains to the expected gainers are worth less than the value of the expected losses to the expected losers.

Second, even if everyone is risk-neutral, there's the problem of not knowing what particular repairs are needed. If your car needs to have the brake pads replaced soon, it's better to know that fact before they wear down too much and you need to replace the rotors too (or worse). And if your car won't start on a cold morning, or overheats on hot days, it's better to know that fact in advance, so you can plan accordingly.

[And if the government's surprise announcement is not a surprise, but is announced well in advance, there's an additional problem. Nobody will change the oil if they know they are going to sell the car anyway. But that's the standard Moral Hazard problem, which is distinct from the Devil You Know problem.]

I wonder which problem matters more, in the used car market. Or in finance?

I started to think about the Devil You Know when thinking about the choice of textbook for Intro Economics. You never really know the flaws in a textbook until you have taught from it. Changing textbooks is always risky; and the flaws catch the teacher unprepared. Most of our switches have gone OK, but one didn't.

Comments

[Pete Bias]

Well, that's enigmatic. Care to elaborate?
Still a fun read as always.

[ltr]

Brilliantly explained.

[Nick Rowe]

Pete and ltr: thanks!

Pete: which was the enigmatic bit? The textbook choice problem? I was reading the CORE textbook online, and thinking about the various costs of switching textbooks, and remembered my old conversation with Mike, and decided to try to analyse the economics of the Better the Devil You Know argument.

[Pete Bias]

Nick, I was looking for exactly which textbooks, not the problem itself. Just curious.

[Mike Smitka]

Of course it's possible to generate information with a textbook, by reading it. This is costly – I'm teaching a new course (Development Economics) this fall, and looked at 4 books. Even if I was able to quickly eliminate 2, it took a while to read enough of the remaining two options to make a choice. And then I had to read the one I chose! Plus I'm reteaching "Macro Principles" for the first time in 9 years. Same issue, but I started with only two books, and made the choice more quickly. Why? One response of publishers is to make rival books as similar as possible, to reduce switching costs. All I scanned was the treatment of the Great Recession, and which of the texts was better on topics international. In the US textbook market, that coverage seems to range from bad to worse. I ended up using Krugman and Wells, not great on the intl side, and it's certainly not as a reflection of the weakness of the authors in that area. Cf. the original Samuelson text from (?) 1947, which had minimal material on trade, simply irrelevant for students in the US at that point in time.

I also just bought a used car. There are many routes, with varying levels of information generated in the search/buying process. At the high end are CPOs (certified pre-owned) at a dealership, which to participate in the program have to follow a long checklist of things from the brand, and typically offer a limited warranty, which makes disinformation even more costly. Then there's checking the history of a vehicle, something that will be absolutely critical after the recent flooding in the US (not so important for Canada!). Where I live, the used car purchaser at a dealership would check whether the car had "lived" above the US snow line, which is also the salt line. In a small community, reputation matters: their used cars wouldn't rust out. Then there's know the owner / know the seller. In my case it was a private transaction where I knew the seller who was the owner. And take a car to an independent garage and pay $!00 or so to have it checked very carefully for problems. Of course you would walk away from any seller who wouldn't let you do this.

One other thing many buyers do is check Consumer Reports and the like. But (whether they realize it or not) that does not get around the Lemons problem, because it is information on the population average, not whether the car you're considering lies above/below it.

The bottom line is that despite the problem of lemons, about 40 million used cars change hands every year in the US plus Canada. I assume the inclusion of Mexico would push that number to 50 million, but have never seen data to know whether the 2:1 used:new sales ratio holds there.

[Economissed]

A few months ago I adopted a cat. He's a nice cat...who turns out to have diabetes and a clotting disorder, not disclosed to me. (He was in "good health except for mild allergies." Hah.)

I plan to update Akerlof's famous paper, this time based on used cats rather than used cars.

(Maybe get in a few nuggets about excess cost growth in the veterinary care sector, too.)

[Iskander]

Nick, I don't think you'd be happy with the CORE textbooks analysis of inflation. Let's just say the money supply isn't directly mentioned (as far as I can remember).

[Nick Rowe]

Pete: it was a very long time ago, and I'm not sure I can remember the particular textbook.

Mike: Welcome to the real cars&economics blogger! I confess I tend not to do enough research when buying used cars. But I usually buy them cheap, and stick with models I know. and things usually work out OK. My guess is that the used/new ratio would be lower for Canada, for the simple reason of rust, so cars don't last as long as in the more southerly US. But that's just a guess.

Economissed: always look a gift horse, or cat, in the mouth?

Iskander: I'm trying to write a post on the CORE text. Yes, the inflation chapter worried me the most. Conflicting claims is a theory of the natural rate of unemployment, not a theory of inflation. But there's some good in the text too.

[Avraam Jack Dectis]

.
The poor kitty!
.

[Phil Koop]

This is a nice example! But while I now understand why the market for lemons is not risk aversion, I still don't understand why it is not moral hazard. I would have thought that market for lemons is a selection effect and failure to maintain is not, but both are examples of moral hazard.

Perhaps you'll need to do another post.

[Nick Rowe]

Phil: thanks! Think of an insurance market. Suppose everyone is identical, and has the same risk of an accident. So there is no adverse selection problem: either everyone or noone buys insurance. But there can still be a moral hazard problem, because if they buy insurance they will make less effort to avoid an accident.

[Tel]

"Making sales compulsory solves the Market for Lemons problem."

Does it??

The way I understand the problem is that if I'm a potential buyer, and I don't know whether the car is a lemon or not, and if there's no way for the seller to convincingly prove this car is not a lemon, then I'm likely to bid low in order to protect myself, therefore making the seller unable to get what would be a fair market price for that car in a situation of perfect knowledge. So now we "solve" this by forcing the seller to dispose of the car at a lower than fair price regardless of her reluctance to sell. I'm not quite sure how this solves anything from the point of view of someone who has looked after their car and is now facing redistribution of property.

What's worse is that since the buyers know it's a forced sale situation, they will bid lower again because suddenly there's a glut of cars on the market and every seller is at a massive disadvantage with the clock ticking.

"The sample of cars offered for sale represents the whole population of cars; those who know their own car is a cream puff can't hold it back from the market. If the Market for Lemons was the only problem in the used car market, then the compulsory universal sale would improve efficiency in the allocation of cars."

Really?!? How do you define "efficiency in the allocation of cars" and how can you demonstrate it has been improved? Who is measuring this? Don't you dare say "Pareto" !

[cm]

"because if they buy insurance they will make less effort to avoid an accident"

Something to consider: Having insurance will also remove one more worry, what happens if something goes wrong, which enables people to act with more confidence and/or more presence of mind. Increased confidence is by far not the same as recklessness!

This also applies not only to confidence/awareness doing activity X when having X insurance - in aggregate when less can go wrong (or something going wrong will likely not have an outsized negative effect on the individual), people may actually perform activities *in other fields* better (because they have less to worry about in aggregate). Not necessarily more carefully, but perhaps focusing on the important stuff and not small things that can go wrong (in the individual's imagination anyway). Again, not in what they are currently doing, but in general.

To turn it around, whatever the situation, would you rather interact with or be helped by somebody whose mind is quite free of worries or somebody who is constantly preoccupied how they may be put into misery due to some small event later today or tomorrow?

[cm]

And it is not only the aspect of the at-fault parties in accidents having insurance - when I can assume that others have adequate insurance, *I* can also be more at ease even though I'm not "planning" to be involved in accidents. Otherwise *I* have to worry (more) about the other party being less cooperative (or fleeing the scene before I can secure evidence), having to go through the legal system and definitely calling the police to the scene even for minor stuff, etc. I was rear-ended several times, and in the latest such incident my car was totaled, but I never called the police because I had confidence in the insurance system (of course also taking the driver and other vehicle information).

The same applies in all other fields. Insurance, especially if mandatory, provides assurance that while adverse outcomes may still occur, at least direct financial damages will be covered, and the problem largely reduces of checking the counterparty's insurance information (or being able to assume with high confidence that adequate insurance exists).

[Nick Rowe]

Tel: imagine that all the prospective buyers of new cars had the option of joining a club. The club rules say that members must sell their cars once a year to other members. If the only problem is the Market for Lemons problem (which is about hidden defects that are not revealed until after you own the car, and not the same as the Moral Hazard problem of defects caused by not looking after the car) all would want to join the club. (I've put them behind the Veil of Ignorance.)

cm: fair points I think.

[Judge Glock]

Doesn't it depend on what you mean by "solving" the Devil You Know Problem? Of course, compulsory sales and purchases detract utility by their very nature, by preventing "non-trades" that are beneficial. Ignoring that aspect, however, isn't the "Devil You Know Problem" just a price adjustment problem? In the compulsory car sale scenario, I think you assume that people must sell their cars at the "average" car value of $500, instead of what might be a risk-adjusted value of, say, $400. Same with the benefits of knowledge about needed repairs. If one factors in in the stable utility of knowing about a car's problems, taking away that knowledge could lower the purchaser's value to, say, $300, even if the average car really has $500 of "objective" value. The loss to purchasers of these utilities is made up by lower prices to sellers. Of course society as a whole losses by the compulsion, but the deals themselves are still beneficial.

I think the textbook market example you use has a similar price adjustment problem at its root. Perhaps the risk-adjusted value to you of adopting a largely unknown textbook is negative $1000. But because textbooks sellers still have limited ability to price discriminate (unlike individual cars, there is supposed to one general price for the "textbook," despite innumerable exceptions in practice), they have problems meeting that price without upending their general market, or going below the product "zero-lower bound" we could say. The possibilities for a mutually advantageous trade, however (you receiving net income for "buying" the textbook for your class, the publisher receiving sales from students), explains the many subsidies textbook publishers offer to professors (I've seen amazon gift cards, free iPads etc.). The Devil You Know Problem then is a problem with any purchase. All prices have to adjust to account for that.

[Nick Rowe]

Judge: "Of course, compulsory sales and purchases detract utility by their very nature, by preventing "non-trades" that are beneficial."

In a large market, compulsory trades are not harmful, if you can sell your car and buy an identical one.

"If one factors in in the stable utility of knowing about a car's problems, taking away that knowledge could lower the purchaser's value to, say, $300, even if the average car really has $500 of "objective" value. The loss to purchasers of these utilities is made up by lower prices to sellers."

If you are selling your car for what it's worth to the buyer, which is less than what it is worth to you, you are worse off, and the buyer is not better off. The trade destroys value.

[Tel]

"In a large market, compulsory trades are not harmful, if you can sell your car and buy an identical one."

But while ever there's a knowledge difference the car you buy cannot possibly be identical. You know about your own car, you don't know exactly what you are buying (even if it does turn out to be physically identical, you still bought the risk). That's the entire basis of the "devil you know" folk wisdom which means every transaction starts out with at least a small loss. Some transactions might bring additional benefit that outweighs the knowledge loss (if I'm leaving the country and not intending to return then probably it's easier to sell the car and worrying about buying another later on). The Judge is quite right, some "non-trades" are beneficial to both parties, and forcing the trade would make both parties worse off.

On reflection, I choose not to join your club.

[Nick Rowe]

Tel: you are repeating the point of my post.

[Tel]

I feel like you are trying to have it both ways, in an attempt to create two distinct problems out of what on the face of it looks like only one problem.

So we agree that an information asymmetry exists, and we agree that a symptom of this is that fewer cars change hands than would do in a world of perfect zero-cost information.

You want to treat the symptom by forcing more cars to change hands, and yet you admit this also causes a certain number of people to be worse off than doing nothing at all. Thus, this might appear to be a partial solution to the information problem, if only we could fixup this darn issue with people being worse off.

Now we have really come full circle, back to the original information asymmetry... forcing people to trade does not create additional information where none existed before. If anything it destroys information which is accumulated by a long term owner and thus making the original problem even worse than to begin with. There's no clear advantage that I can see in compartmentalizing the gains from forced trade (some parties will gain) and isolating that from the losses (other parties will lose but you want to call that a different problem).

Let me try an analogy: suppose one drug has two effects and one of those effects is beneficial while the other is harmful. Now that's just a basic trade off which only the individual can judge under the circumstances (in some situations, better not to take that drug at all). But suppose we call the good effect "a cure" and the bad effect "a side effect" and now we can say, "Hey this drug will cure you but there may be a few side effects"... it's just a different way of looking at the same thing, the emphasis has shifted, the reality is unmoved.

[Frank Restly]

Nick,

"The sample of cars offered for sale represents the whole population of cars; those who know their own car is a cream puff can't hold it back from the market."

Perhaps not, but they can change their cream puff cars into lemons by stripping off / replacing parts to gain a competitive advantage.

"Making sales compulsory solves the Market for Lemons problem."

I don't think it does. Compulsory sales invites a race to the bottom where each person strips down / reworks his or her car to give the least value to the next person.

Instead, the solution to the Market for Lemons problem is to eliminate the information asymmetry - all cars are thoroughly inspected by an independent third party (government) prior to sales being finalized.

Or if government intervention is not your thing, tri-party sales are another way to go. In this arrangement, every sale consists of Joe selling his car to Jane who sells her car to Bill who sells his car to Joe. Any sale must be approved by two of the three parties to go through.

And so each bidder prices every car (excluding their own) and the highest value "triangles" are then created from the group of bids.
With a group of 6 cars for sale, there will be two tri-party sales comprised from a list of 10 possible sale combinations.
If the sellers are Joe, Jane, Bill, Roger, Rachel, and Tom, then the 10 possible combinations are as follows:

Joe, Jane, Bill (Sale #1) and Roger, Rachel, Tom (Sale #2)
Joe, Jane, Roger (Sale #1) and Bill, Rachel, Tom (Sale #2)
Joe, Jane, Rachel (Sale #1) and Bill, Roger, Tom (Sale #2)
Joe, Jane, Tom (Sale #1) and Bill, Roger, Rachel (Sale #2)
Joe, Bill, Roger (Sale #1) and Jane, Rachel, Tom (Sale #2)
Joe, Bill, Rachel (Sale #1) and Bill, Roger, Tom (Sale #2)
Joe, Bill, Tom (Sale #1) and Jane, Roger, Rachel (Sale #2)
Joe, Roger, Rachel (Sale #1) and Jane, Bill, Tom (Sale #2)
Joe, Roger, Tom (Sale #1) and Jane, Bill, Rachel (Sale #2)
Joe, Rachel, Tom (Sale #1) and Jane, Bill, Roger (Sale #2)

With a 2/3 majority vote required for all sales, the incentive to "lemonize" your own car is reduced.
The sale may not get finalized and you are left with putting your own car back together.

See more on Tri-Party agreements here:
http://www.investopedia.com/terms/t/tri-party-agreement.asp

[Frank Restly]

More to the above.

With no prevailing voting strategy and a 2/3 majority required, 50% of sales will go through and the other 50% will fail.
The possible vote combinations are:
YYY
YYN
YNY
NYY
YNN
NYN
NNY
NNN

The four reasonable voting strategies would be based upon
1. Did a get a good price on the car I am selling?
2. Did I get good value on the car I am buying?

Strategy #1 - I vote yes if both conditions are satisfied (25% of the time I vote yes)
Strategy #2 - I vote yes if I get a good price on the car I am selling (50% of the time I vote yes)
Strategy #3 - I vote yes if I get good value on the car I am buying (50% of the time I vote yes)
Strategy #4 - I vote yes if I either get good value on the car I am buying or I get a good price on the car I am selling (75% of the time I vote yes)

In a multi-turn learned strategy game, a rash of fraudulent lemons (as opposed to advertised lemons) should cause fewer players to adopt aggressive strategy #4 and more players to adopt conservative strategy #1. If all players adopt Strategy #1, then only about 25% x 50% = 12.5% (1 sale in 8) ever go through.

[Frank Restly]

Small math mistake:

If all players adopt voting Strategy #1, then only about (1/4 * 1/4 * 1/4 + 1/4 * 1/4 * 3/4 + 1/4 * 3/4 * 1/4 + 3/4 * 1/4 * 1/4) = 15.625% (10/64) of sales go through.
On the other hand, if all players adopt voting Strategy #4, then (3/4 * 3/4 * 3/4 + 3/4 * 3/4 * 1/4 + 3/4 * 1/4 * 3/4 +1/4 * 3/4 * 3/4) = 84.375% (54/64) of sales go through.

Consider if we change to matched pair buyers and sellers (instead of tri-party sales)
On the upside, with aggressive voters (Strategy #4) and a single yes vote, fully (3/4 * 3/4 + 1/4 * 3/4 + 3/4 * 1/4) = 93.75% (15/16) of sales go through.
However, even if voters turn conservative (Strategy #1) and a single yes vote gets the sale made, (1/4 * 1/4 + 3/4 * 1/4 + 3/4 * 1/4) = 43.75% (7/16) of sales go through.

Yes, you will get more sales made in a matched pair sales arrangement when used car buyers / sellers become aggressive, but you will likely get too many used car sales when used car buyers / sellers become conservative.

[Tel]

"Compulsory sales invites a race to the bottom where each person strips down / reworks his or her car to give the least value to the next person."

At least you have removed the information asymmetry by that method... everyone knows they always buy a lemon. The price will then adjust to the absolute minimum that can get any driver to the end of the year when forced sale hits them again.

A few left wingers might start complaining about the whole disposable society thing, but we can safely ignore those loonies. They don't understand capital investment.

[louis]

Frank Restly: "Instead, the solution to the Market for Lemons problem is to eliminate the information asymmetry - all cars are thoroughly inspected by an independent third party (government) prior to sales being finalized."

You could probably get the same result by requiring sellers to warranty the condition of the cars they sell and making them liable after the fact for any defects.

[Frank Restly]

Tel,

"At least you have removed the information asymmetry by that method... everyone knows they always buy a lemon."

Perhaps, but I think we can agree that while the total volume (quantity of cars) trade may increase with compulsory sales, the total value (quantity of cars x price per car) will likely decrease.

Louis,

"You could probably get the same result by requiring sellers to warranty the condition of the cars they sell and making them liable after the fact for any defects."

If government (or some other party) is responsible for enforcing warrantees, they are left to determine whether a reported defect occurred prior to or after the sale is made. Sure there is more upfront effort in inspecting cars prior to them being sold, but there would be fewer he said / she said arguments after the fact.

[louis]

"If government (or some other party) is responsible for enforcing warrantees, they are left to determine whether a reported defect occurred prior to or after the sale is made. Sure there is more upfront effort in inspecting cars prior to them being sold, but there would be fewer he said / she said arguments after the fact."

Right approach would be circumstances based. If it's easy to identify defects pre-sale, your method works. If it's hard, then maybe the right approach is strict liability within a certain timeframe (6 months, 1 yr, whatever) where any defect is assumed to be attributable to a flaw knowable before sale. The "false positive" error would just get baked into price at sale.

[Frank Restly]

Louis. Agreed.

Any thoughts on tri-party trade arrangements?
In my mind, it represents a compromise on Nick's two party compulsory trades.

The risk / reward payoff for fraudulently representing a car that you have "lemonized" falls off under tri-party sales.

With two party sales, you are stuck between two suboptimal choices - one yes vote required to approve a trade(Fraudster's Paradise) or two yes votes required to approve a trade (Basically what we have now - inefficient allocation of used cars).

Notice that tri-party trade arrangements do not rely on increased information flow.
I don't need to know the car that I am buying as well as my own.
All that has happened is that the incentive for providing false information have been reduced.

[Frank Restly]

More on tri-party arrangements (in this case tri-party repo) here:

http://libertystreeteconomics.newyorkfed.org/2015/05/financial-innovation-the-origins-of-the-tri-party-repo-market.html
http://libertystreeteconomics.newyorkfed.org/2015/05/financial-innovation-evolution-of-the-tri-party-repo-arrangement.html

The premise:

"In the tri-party repo arrangement, Salomon—the dealer—and all its potential repo lenders had accounts on the books of the clearing bank, Manufacturers Hanover. Settlement would occur by Manufacturers Hanover transferring cash and securities between the accounts of Salomon and Salomon’s lenders on Manufacturers Hanover's books. Under this arrangement, the dealer no longer depended on Fedwire to receive the securities from one lender and to send them to the other lender’s account. Indeed, settlement could occur at any time, even after the close of Fedwire Securities."

Notice that in this instance, Manufacturer's Hanover absorbs none of the risk (and has no up / down voting stake) in the trade. They are simply a clearinghouse (replacing Fedwire) between Solomon Brothers and it's potential lenders. That is a non-issue when the collateral being used is Treasuries. It becomes a big issue when risky securities (equities, MBS, corporate bonds) are offered as collateral.

This is different than what I am describing above.
All three parties should have a voting stake and all three parties should bear risk.

[louis]

Frank - I read through your tri-party sales comments and I don't think I get it.
The core issue in mkt for lemons is assymetric info. Seller knows something the buyer doesn't, the buyer can't easily uncover this information before purchase, and the seller can't easily and verifiably provide this information to the buyer.
I'm not sure how introducing another layer of purchasers changes the core assymetric info problem, but maybe I'm missing something.

[Frank Restly]

Louis,

"The core issue in mkt for lemons is asymmetric info. Seller knows something the buyer doesn't, the buyer can't easily uncover this information before purchase, and the seller can't easily and verifiably provide this information to the buyer."

See my above:

"Notice that tri-party trade arrangements do not rely on increased information flow. I don't need to know the car that I am buying as well as my own. All that has happened is that the incentive for providing false information has been reduced."

And that is the important part - the incentive for providing false information has been reduced.
If the seller is offering a lemon for sale, tells me that the car is a lemon, and tells me what exactly is wrong with it - then there is no issue. I can adjust the price I offer for the car accordingly.

What we are concerned about is information that the owner refuses to divulge either through ignorance or deceit - yes?
That information could be obtained through pre-sale inspections / after-sale warrantees as you and I discuss above - yes?

But instead of worrying about the actual information that the seller is trying to hide / doesn't know,
can't we instead focus on limiting the success rate of him selling his lemon to the unsuspecting buyer?

That is why I went with tri-party sales. For the sale to go through, 2 of the 3 car owners must vote yes to approve the sale.
With no prevalent voting strategy, 3 people will reach a 2/3rd's majority only 50% of the time. This is an improvement over Nick's compulsory sales between two people - each sale goes through 100% of the time.

[louis]

Ok so 3 parties to the sale. Two have good cars, one has a lemon. The lemon is undetectable.
What prevents the trade from happening? Where is the improvement?
In fact I see the same market degeneration happening here as in the classic example. If you own a good car and participate in the trade, at best you are left in the same place, and there's a real possibility you are worse off. So if you have a good car, no incentive to trade. Everyone knows this, and everyone knows everyone knows this, so people only offer to trade lemons.

[Frank Restly]

Louis,

"Ok so 3 parties to the sale. Two have good cars, one has a lemon. The lemon is undetectable."
"What prevents the trade from happening?"

The owner of the lemon knows that he has a lemon, but does not reveal he has a lemon. He is a yes vote. If the other two owners vote no - then the sale doesn't go through. If one or both of the two other owners votes yes, then the sale does go through.

On a straight up vote, it takes a 2/3 majority vote from the 3 participants to get the sale to go through. See my voting matrix above. With no prevailing voting strategy, about 50% of trades will go through, 50% of trades won't go through.