« Democrats Proudly Cut Medicare Benefits | Main | What I've Been Reading »
More evidence against the lemons model
...the empirical analysis finds that cars that were acquired used required no more maintenance expenditures than those that were acquired new of a similar age.
Here is the link, here is the published version, hat tip is from Division of Labor, and here is Alex's previous post on adverse selection more generally.
Posted by Tyler Cowen on July 21, 2008 at 10:55 AM in Economics | Permalink
Comments
But isn't there a censoring problem, in that markets affected by the lemons problem don't exist, while those that do, have solved the problem?
Posted by: Eric at Jul 21, 2008 11:22:24 AM
I also find it surprising that he controlled for age and not mileage, and yet there was still no difference.
By the way, what the heck are beta_2, beta_3, beta_4, and epsilon in equation 1? He references a paper by Bond, but I couldn't find that formula anywhere in the Bond paper. How does he solve for one variable with 1 equation and 5 unknowns? (For the record, I have absolutely no formal training in economics, so forgive my naivety.)
Posted by: Michael at Jul 21, 2008 11:32:20 AM
>> How does he solve for one variable with 1 equation and 5 unknowns?
That is the magic of economerics. You are better off not knowing it.
Posted by: bb at Jul 21, 2008 11:44:24 AM
Eric: Yes, but only if the limit argument holds. We can rule out the limit argument with casual observation, though: I bought my car used from a stranger (and it's just great 5 years later, thanks for asking).
Michael: Using regression analysis.
Posted by: o at Jul 21, 2008 12:47:23 PM
The problem with used car pricing is as much one of information asymmetry as one of actual quality. The seller knows if his car has a cracked frame and the buyer doesn't. The seller knows the true mileage (if it was dialed back) and the buyer doesn't. The distribution of problems and mileage may well be identical between the cars purchased used and those purchased new of the same age. But the used car may have issues or mileage unknown to the buyer. That's why a car loses so much value as soon as it is driven off the lot. In fact, a buyer of a car that's one month old may well have more cause to be suspicious.
Posted by: Max Rockbin at Jul 21, 2008 12:51:14 PM
I wonder if it always makes sense to buy a used car. I know some people swear by it, but is it always the best choice.
For example. A new 2008 Honda Accord LX is 18,458. A 2006 Honda Accord LX is 13,774. So, buying used would save you 4,684. Now, an extended warranty to match the lost 2 years of the Honda Warranty could cost upwards of $1,500. Taking the savings down to 3,184. Also, with the new car you can get 4,99% financing from Honda for 60 months. That would save you $1800. That takes the savings down to $1384.
Now, obviously a 08 Accord has two more years of life in it than a 06 Accord. Is that savings of 1384 really enough to warrant giving up 2 years of vehicle life?
It doesn't seem that way to me...
People see the new price of 18,458 and the used price of 12,774 and say see I "saved" 4,684. But the reality is they saved far less...
Posted by: jmo at Jul 21, 2008 1:13:50 PM
Unless of course the warranty is an unnecessary expense.
What are the odds, after all, that a nearly new Honda will need to go to the shop in two years?
It would be worth it for some people to gamble 4684 or even 1500 against the probability that an 06 Accord would have non-routine maintenance expense greater than the savings in a 2 year period.
Posted by: toxic at Jul 21, 2008 1:19:55 PM
Not really evidence against the lemons model is it? The abstract suggests that leasing and CPO are solutions that have arisen to solve the lemons problem. Isn't this more or less in accord with the basic idea that costly signalling is a way out?
Posted by: Bernard Yomtov at Jul 21, 2008 1:57:00 PM
Obviously CPOs are one way to deal with the lemon problem, but why would leasing have anything to do with the lemon problem?
In my life I have bought four cars from rental companies and have been quite satisfied with every one. By the way they wash rental cars every time they are turned in so they are probably washed more often than other cars.
Posted by: spencer at Jul 21, 2008 2:09:21 PM
Maybe it's just that cars are so good today, and warranties so comprehensive, that there isn't anything like a "lemon" any more. If a car has problems, the part gets replaced, and, in future, is no more likely to break down than any other car.
Suppose you own a car that's still under warranty. What could be happening with it that would convince you that it would be less reliable in future, keeping in mind that if there's something wrong with it NOW, you can get it fixed free?
Posted by: Phil Birnbaum at Jul 21, 2008 2:22:43 PM
Phil - If the car was in accident. I had a car that got t-boned ahead of the front wheel and the fix required removing the engine. The body shop did a piss poor job. They didn't attach a coolant hose properly and it leaked all its coolant. They also overfilled the A/C causing it to fail. They also didn't seem to do a good job reattaching the electrical as some gremlins popped up.
Posted by: jmo at Jul 21, 2008 2:43:24 PM
jmo: agreed, that's the only one I could think of too. But when I bought my used car, the dealer certified that it hadn't been in an accident (at my request). That seems like something that's easy for the market to take care of.
Posted by: Phil Birnbaum at Jul 21, 2008 3:06:02 PM
Phil, The other concern I would have is a marjor component failure. Given the issues the body shop had with taking out the engine and putting it back in - if a car serious engine damage, due to timing belt failure, coolant loss, loss of oil etc, oil sludge, etc. and it required major repair work I would worry. I can only imagine the billions of dollars and countless hours that went into the process enginering that ensures the Accords are manufactured properly at the factory. I would imagine the same attention to detail isn't present at your average Honda dealer service department.
Posted by: Jmo at Jul 21, 2008 3:22:47 PM
jmo,
What do you mean by this: "Also, with the new car you can get 4,99% financing from Honda for 60 months. That would save you $1800." How does paying interest save you money? I would understand if it was 0% (or below the rate paid by banks on deposits), but otherwise I fail to see how you could possibly make a 10% return by borrowing $18k for 5 years at 5%. Are you investing this money in the stock market? Gold? There is no end to the absurdities you can justify by assuming a high enough rate of return on borrowed funds.
Are you comparing the loan to an auto loan from a bank? Similar rates can be obtained from credit unions. You would have to have quite a rate difference to save 10% of the value of the car over 5 years. What is your benchmark, a 25% loan?
I suspect you are biased.
Posted by: Cliff at Jul 21, 2008 4:10:29 PM
Gee. $31.50 to buy a copy.
As the man in the Miller commercial says, "You people must be crazy."
Wonder how many they sell.
Posted by: Bernard Yomtov at Jul 21, 2008 4:21:05 PM
Cliff, 60 month car loan at 8.5% = $380 a month for 60 months vs. 4.99 = 348.24. savings of 31.76 a month 31.76*60 = 1905.60. Being able to borrow @ 4.99 vs. 8.5 = a savings of 1905 over 60 months.
Posted by: jmo at Jul 21, 2008 4:25:42 PM
In Venezuela used car from the 70 and 80s are sold at higher price than new cars.
A 79 malibu a " regulated" car that is a car with only the basic with a price fixed by the government were until 2 or 3 years ago the most valued car in the country.Ltd ( 78 or older), Caprice 80 and 81 ( the 80 model is more expensive than the 81)and Fairline 78 were used for passenger transportation until the government banned them in 2004. A 1990 corolla sells by 1/3 of the value of a 2008 model
Posted by: k at Jul 21, 2008 4:44:24 PM
This isn't evidence against the lemon model - it is simply stating that, under current market conditions a market failure doesn't occur.
I'm pretty sure that back when I took first year economics I learn't that as long as the value the seller puts on their "good quality" car is below the value a buyer is willing to pay for a car of unknown quality, then a market with both good and poor quality cars would be sustainable - this is still consistent with the lemons model.
Posted by: Matt Nolan at Jul 21, 2008 5:17:01 PM
The question he asks cannot, no matter what answer he gets, contradict the lemons model. In fact, one could as well argue that his econometric result is evidence in favor of the lemons model.
Reason:
1.) Under a severe lemons problem, it is easy to imagine, that the only used cars that are sold are the impeccable ones. If your car has even the most minor defect you cannot sell. Then the maintenance expenditures will be much lower for used cars.
2.) People have a different propensity to sell their car. There are people who get a new car every year, no matter what. Maybe these people also treat their cars differently.
Posted by: Martin at Jul 21, 2008 9:41:54 PM
"Cliff, 60 month car loan at 8.5% = $380 a month for 60 months vs. 4.99 = 348.24. savings of 31.76 a month 31.76*60 = 1905.60. Being able to borrow @ 4.99 vs. 8.5 = a savings of 1905 over 60 months."
JMO, you've got this all wrong. Assume, as you seem to be, that you're putting no money down. (I have no idea why you'd assume this, but for the sake of argument, I'll go with it.)
Your interest cost over five years on the used car -- borrowing $13,774 at 8.5% -- is $5854. Your interest cost over five years on the new car -- borrowing $18,458 at 4.99% -- is $4605. So your interest savings are $1200, not $1900. On top of that, you need to factor in the interest you earn on the extra $1000 in principal you're not paying every year. Let's say you put it in the stock market and make 7% annually, giving you around $400, reducing total interest savings to $800. Regardless, it doesn't come close to making up for the extra $5000 you don't have in your pocket. And anyone who shells out $1500 for a two-year warranty on a Honda is a sucker.
I also don't know why you're assuming you can't get cheap financing for used cars. I just bought a very nice certified used car and got 1.9% financing for 60 months. There are plenty of offers like that out there.
Posted by: K. Williams at Jul 21, 2008 11:50:35 PM
Didn't read the paper since I'm not going to blow money on it. It occurs to me, however, that the same folks who are likely to buy a new a car are also likely to have uneconomic repairs done.
For instance, people like me who typically buy 6-8 year old vehicles with 100K miles in a model with a good CU repair record, are far less likely to have things fixed which are minor inconveniences. Whereas, a party who bought that car new but does not sell it, seems far more likely to have that ding taken out of the door, the weatherstripping replaced, the cracked dashboard replaced, the transmission that has to be babied between 1st and 2nd replaced, etc. You are comparing the repair records for two populations where the car is not the only differentiated variable, the car owner is self-selected as either free-spending (new car buyer) or value-driven (used car purchaser).
Posted by: benamery21 at Jul 22, 2008 12:22:45 AM
What Phil Burnbaum said - true lemons, even in Detroit cars, are much rarer now than when Akerlof wrote. And factory warranties are far longer too (with fewer lemons they can afford to be), so they'll cover late model used cars.
So you're less likely to bite on a lemon and if you do it won't taste as bitter as it once did. It's possible that the market failure once existed but no longer does.
Anyway, even if Akerlof was empirically wrong about used car markets this wouldn't change the tremendous fruitfulness (sic) of his insight, which has important applications in fields from finance to labour economics. Plus his short article is a model of clear exposition and exploration of an idea.
Posted by: derrida derider at Jul 22, 2008 1:24:02 AM
Just out of interest, does anyone know of an empirical study that does show a real life example of the lemon model, i.e. high quality products are sold for the same price as low quality products because differentiation is hard?
Posted by: Great Zamfir at Jul 22, 2008 7:25:04 AM
There seems to be one question I've never seen an economist ask in response to the Akerlof's original paper: Are new cars overpriced and used cars actually an accurate reflection of a vehicle's value?
Posted by: Ted Craig at Jul 22, 2008 9:28:58 AM
Bvlgari Jewelry
Replica Bvlgari Jewelry
Bvlgari Replica Jewelry
Gucci Jewelry
Replica Gucci Jewelry
Posted by: aion kina at Mar 18, 2009 9:08:52 PM
