May 14, 2006

Markets in Imperfect Information - Lemons, Limes and Silver Bullets

Twan points to a nice slate/FT article on the market for lemons:

In 1966 an assistant economics professor, George Akerlof, tried to explain why this is so in a working paper called "The Market for 'Lemons.'" His basic insight was simple: If somebody who has plenty of experience driving a particular car is keen to sell it to you, why should you be so keen to buy it?

Akerlof showed that insight could have dramatic consequences. Buyers' perfectly sensible fears of being ripped off could, in principle, wipe out the entire used-car market: There would be no price that a rational seller would offer that was low enough to make the sale. The deeper the discount, the more the buyer would be sure that the car was a terrible lemon.

If you are unfamiliar with Akerlof's market for lemons, you should read that article in full, and then come back.

This whole area of lemons is sometimes called markets in asymmetric information - as the seller of the car has the information that you the buyer doesn't. Of course, asymmetries can go both ways, and sometimes you have the information whereas the other guy, the seller, does not. What's up with that?

Well, it means that you won't be able to get a good deal, either. This is the market in insurance, as described in the article, and also the market in taxation. These areas were covered by Mirlees in 1970, and Rothschild & Stiglitz in 1976. For sake of differentiation, as sometimes these details matter, I call this the market for limes.

But there is one final space. What happens when neither party knows the good they are buying?

Our gut reaction might be that these markets can't exist, but Michael Spence says they do. His example was the market for education, specifically degrees. In his 1973 paper entitled "Job Market Signalling" he described how the market for education and jobs was stable in the presence of signals that had no bearing on what the nominal goal was. That is, if the market believed a degree in arts was necessary for a job, then that's what they looked for. Likewise, and he covers this, if the market believed that being male was needed for a job, then that belief was also stable - something that cuts right to the core of our beliefs, because such a belief is indeed generally irrelevant but stable, whether we like it or not.

This one I term the Market for Silver Bullets, a term of art in the computing field for a product that is believed to solve everything. I came to this conclusion after researching the market for security, and discovering that security is a good in Spence's space, not in Akerlof's nor Rothschild and Stiglitz's spaces. That is, security is not in the market for lemons nor limes - it's in the tasteless spot in the bottom right hand.

Yup, because it is economics, we must have a two by two diagram:

The Market for Goods,
as described by Information
and by Party
Efficient GoodsLemons
(used cars)
(Tax, Insurance)
Silver Bullets

Figure 1. Security is a Symmetrically Insufficient Market

Michael Spence coined and explored the sense of signals as being proxies for the information that parties were seeking. In his model, a degree was a signal, that may or may not reveal something of use. But it's a signal because we all agree that we want it.

Unfortunately, many people - both economists and people outside the field - have conflated all these markets and thus been lead down the garden path in their search for fruit. Spence's market in silver bullets is not the same thing as Akerlof's market in lemons. The former has signals, the latter does not. The latter has institutions, the former does not. To get the full picture here we need to actually do some hard work like read the original source papers mentioned above (Akerlof and Spence aren't so bad, but Rothschild & Stiglitz were tougher. I've not yet tried Mirrlees, and I got bogged down in Vickery. All of these require a trip to the library, as they are well-pre-net papers.)

In particular, and I expand on this in a working draft paper, the bitter-sweet truth is that the market for security is a market for silver bullets. This has profound implications for security research. But for those, you'll have to read the paper :)

Posted by iang at May 14, 2006 04:32 PM | TrackBack

The economics of "Keeping the Game Going". If two sides to a game derive pleasure and comfort they will seek to tell lies and engage in fruitless activities to keep the game going. The game can be a contract or the illusion of acquiring a new or previously owned car. If anyone has been in a contractual relationship like marriage that ended in divorce they have already experienced this situation its called for Old Time Sake and security professionals at organizations buy products from friends that worked as security professionals previously. That fact is regardless of the institutional framework deals are personal and experience especially on a long term basis all the depth of another relationship. The young new upstart that knows new software and silver bullets displaces the old upstart and his array of relationships promising a silver bullet. The information that is known is the relationship between the two people of the people in the group that are empowered to contractually obligate each other. Why should I buy anything from anyone based on product information when product information is not what drives 99% of what I live for. I live for people. The economics of my relationships is derived as a secondary reality and can only reflect aspects of my emotions. Why do the children of the rich and powerful tend to become rich and powerful? The answer is because Daddy thinks his children are worthy to his wealth. What things do children do to make Daddy happy? They tend to do what Daddy has done because he understands that. Why would I buy a used car? The answer is economic in that my ability to buy a new one is not up to snuff and the need is greater than the risk of a lemon. I can buy a cheap used car and owe the mechanic the money to fix it at a lower cost than the note on a new car or a better quality used car with higher interest rates. Mechanics are local and depend on the sale of petrol and other items in their store so interest rates are reduced. In order to obtain an inspection I require a mechanic to alter the true results so I can pass and drive. I buy used cars because the unofficial economy permits the transaction to happen in cash avoiding a tax audit. I buy used cars because new cars if they are lemons cost much more to fix with dealer rates in the garage.
I buy security software because it is Government rated. I buy silver bullets because my peers will think I'm cool. I buy silver bullets because the vendor has a backend deal and I need the vacation. I buy silver bullets because my personal network after the purge of outsourcing leaves me no avenue to increase its economic viability as a network therefore I'm threatened. I buy silver bullets because I have purchased them in the past and there is no accounting for my poor purchases. I buy silver bullets because I know nothing but neither does anyone else and if they do we do not hire them because they will end the game the dirty bastards.

Posted by: Jimbo at May 14, 2006 07:29 AM

part of this drifts into the area when something appears to be valuable ... and there are items that are taken as representations of that something ... then you have people inclined to fraud. in various of my postings on credentials, certificates, diplomas, licenses, etc that have served for centuries as representation paradigm (for replying parties that have no direct knowledge and/or means of obtaining that knowledge) ... education has become a valuable commodity and diplomas that are taken as representation of education, are now being fabricated (either fraudulent and/or in diploma mills).

this is the scenario that there are various facts ... and for centuries; credentials, certificates, diplomas, and licenses have been used to represent various facts for relying parties that have not the means of directly accessing the facts themselves. with advances in dataprocessing and online networks there are fewer and fewer circumstances where relying parties can't find some means of accessing the facts ... drastically reducing the situations where separate and distinct credentials, certificates, diplomas, and/or licenses are required (as a means of representing such facts).

this is sort of the certificate sticker on the window of a used car vis-a-vis getting something like a carfax report.

misc. past posts mentioning the centuries old paradigm of credentials, certificates, diplomas and/or licenses for relying parties that don't have any other means of accessing the information Separation of Roles - an example History and definition of the term 'principal'? Court rules email addresses are not signatures, and signs death warrant for Digital Signatures Is The Public Key Infrastructure Outdated? X.509 and ssh X.509 and ssh X.509 and ssh X.509 and ssh X.509 and ssh X.509 and ssh confidence in CA Multi-layered PKI implementation

Posted by: Lynn Wheeler at May 15, 2006 11:27 AM

Actually Wikipedia on Spence says:

Michael Spence is probably most famous for his job-market signaling model, which essentially triggered the enormous literature in this branch of contract theory. In this model, employees signal their respective skills to employers by acquiring a certain degree of education, which is costly to them. Employers will pay higher wages to more educated employees, because they know that the proportion of employees with high abilities is higher among the educated ones, as it is less costly for them to acquire education than it is for employees with low abilities. For the model to work, it is not even necessary for education to have any intrinsic value if it can convey information about the sender (employee) to the recipient (employer) and if the signal is costly.

Which is a better description.

Posted by: Wikipedia on Spence at April 19, 2007 01:51 PM

The silver bullet market fits right into the theory of evolutionary sexual selection. (Typically) females want to pick healthy, robust males. They have no direct way to test that, so the males evolve signals of health and robustness. To be useful as signals, they have to be expensive - it must be difficult for unhealthy, non-robust males to send the signals. Once the signals become established, they take on a life of their own - resulting in such famous examples as the peacock's tail.

It's not as if we humans are immune, as expensive cars and other displays of wealth make clear.

-- Jerry

Posted by: Jerry Leichter at June 4, 2014 05:16 PM
Post a comment

Remember personal info?

Hit preview to see your comment as it would be displayed.