Sympathetic Post-Keynesian types often ask me what I think of the whole ‘asymmetric information’ paradigm. They’re often struck when I say that I think that its vacuous. After all doesn’t this paradigm undermine the dreaded General Equilibrium theory? Well yes it does but that doesn’t mean that it is in any way substantive. You can’t just lend a paradigm credence because it produces results that overlap with your own.
In his book Minding the Markets — about which I will be writing more posts on in the coming days — David Tuckett provides a very nice summary of George Akerflof’s famous ‘market for lemons’ paper. I actually did this in my masters degree — marginalist microeconomists love this paper because of its Sudoku-puzzle qualities — but I never wanted to write about it because I didn’t want to bother summarising it. Tuckett has done me a great service in this regard. He writes:
Akerlof supposed that in the secondhand car market well- informed sellers face ignorant buyers and that there were two kinds of car – reliable cars and lemons. The seller knows which he thinks he has but it is difficult for the buyer to tell. His formal analysis showed how the price of used cars will be discounted to reflect the incidence of lemons in the population. It will be an average of the values of good cars and lemons. But that average is a good price for the owner of a lemon, but a disappointing price for the seller of a reliable car. So owners of lemons will want to sell and owners of reliable cars will not. As buyers discover this, that knowledge will pull down the price of secondhand cars. And things will get worse. The lower the average price, the more reluctant the owners of more reliable cars will be to sell and the more suspicious buyers will get, driving things down further. The end result will be that secondhand cars will be of poor quality and many secondhand cars will be bad buys even at low prices. (p5)
That’s basically it. That is pretty much what all the fuss is about. The student is then taught to work this out carefully in some sort of tedious problem set form. The exercise shows that within the given assumptions the market breaks down pretty much completely. Akerlof provisionally reaches pretty absurd conclusions when he writes:
The cost of dishonesty, therefore, lies not only in the amount by which the purchaser is cheated; the cost also must include the loss incurred from driving legitimate business out of existence.
Really? People being dishonest in a market drive honest people out of this market? Really? Does the author genuinely think that this is a description of the real world? Well, in the real world there are many car dealers that sell junk and yet, for some bizarre reason, the market continues to exist.
So, why then doesn’t the market break down? Well, anyone who has ever dealt in second-hand cars — I used to buy and sell them when I was a teenager — knows that this is all a question of trust. You must convince the customers that the car is not a lemon. You must provide papers documenting the car’s service history and so forth. Maybe they will even hire a mechanic to test the car. Basically, the seller who has limited information tries to gain more information. (Actually, in my experience the seller often doesn’t have full information about the car either if they are a dealer/arbitrager, but I digress…).
Alternatively, the dealer could just try to talk the customer into buying a piece of crap. This happens too, so it can’t be ruled out. But it sure as hell doesn’t lead to the market imploding. Anyway, leaving this aside, Akerlof basically notes this. Tuckett points this out when he writes:
Since we have markets, the conclusion highlights how building trust must be a crucial element in the way financial markets work and demonstrates how parsimonious abstract modelling can very efficiently and rigorously get to the heart of a matter. Buyers can only be persuaded to trust sellers and so come into the market if the underlying situation of information asymmetry is somehow modified. One way is for sellers to try to frame the information context in which decisions are made to make the buyer more confident in the seller – for example, by advertising ‘one owner’ or ‘lady driver’, by offering to show service records or a report from an independent agency, or by taking explicit measures to share the risk of things going wrong in future, such as a guarantee from a reputable source. Some of these devices are discussed in Akerlof’s original paper. (pp5-6)
Okay great. So why do I say that this is vacuous? Well, think about how this argument is fobbed off on the student. First, the student is told to assume perfect information. They are told to think about a very abstract model in which everyone knows everything and prices reflect this perfectly. When the critical student says that they do not believe how this is real markets work they are told to shut up and do their problem set. When they bring up anecdotal evidence that markets work completely differently and have much to do with power-relations, emotions and trust they are again told to shut up.
So, the student either shuts up or they migrate to a more realistic discipline. If they shut up for long enough they eventually move on to intermediate microeconomics. Now they are told to take on board a few instances in which issues such as trust arise. They are then showed how this undermines the perfect information model that they spent a number of years working their asses off to master. Now they are told that these more psychological or sociological aspects of markets need to be studied.
Well then what the hell was the point of studying the perfect markets stuff in the first place!? Any reasonably bright 18 year old can see that markets are predominantly determined by issues of power, emotion and trust. So, why didn’t we just run with it in the first place?
All Akerlof’s model shows is that if you don’t assume perfect information but you do assume some vague notion of rationality the market will break down. Then he provides a few hand-waves as to what actually stabilises markets — which are the very factors that students are told not to think about when they start studying markets! Students are basically led around on a wild goose chase and come out the other side with a bunch of tools that are altogether useless for studying real world markets.
The end result is that the students will try to model these issues of power/emotion/trust using the very same tools that were only really appropriate for the perfect information, perfect rationality assumptions. There’s a term for this in the philosophy of science: its called a ‘degenerating research program’. A research program is degenerative when ad hoc additions are made not to produce novel facts but to defend core assumptions. In this case the core assumptions are those of rationality and market-clearing prices and so forth, while the ad hoc assumptions are handwaves toward issues of trust and so forth that cannot really be studied using the tools that practitioners spend years of their lives acquiring.
A progressive research program would start from the assumption that issues like trust, emotions and power play a fundamental role and would then build on this. So, for example, a progressive researcher might look into existing literature on the psychology and sociology of trust and then try to integrate this with basic economic insights about markets in the study of any particular markets which they would approach not deductively but empirically. This is, in fact, what Tuckett is doing in the book but I will leave this for another day. For now it should just be made clear that at best the asymmetric information paradigm is a degenerating research program, at worst it is a vacuous waste of time and money which is turning out useless economic students that spend their whole time chasing their own tails and have no capacity to actually study interesting multidisciplinary approaches to microeconomics.
Finally, a note on intent. Akerlof warmly received Tuckett’s book and gave it the highest of praise. But Tuckett is completely shunning the core assumptions of marginalist microeconomics. Are we then to assume that Akerlof’s intent in his famous paper was wholly destructive? Should we infer that the idea was to tear down the General Equilibrium edifice so that people could take more realistic approaches? I am certainly willing to entertain the possibility. But this is not what happens. Rather students continue to chase wild geese and learn mathematical tools that are completely inappropriate for dealing with the problems they are studying. They then try to apply these tools to the material and they end up with theories that are just as arcane and irrelevant to the real world as are the old perfect information General Equilibrium models. If you want to tear up the old paradigm you have to do so at the roots. Attacking the branches does nothing at all. In fact, it strengthens it.
Fixing the Economists 
This is very interesting, because I’m a big fan of William K. Black’s frequent use of the Gresham’s Dynamic, which he writes about a lot over on NEP, and he seems to base that off of George Akerlof’s “Market for Lemon’s” example there, and the general idea of asymmetric information.
I’ve actually found the Gresham’s Dynamic to be one of the most interesting/useful things I’ve learned about, for getting perspective on some economic issues surrounding fraud (it can be applied to an extremely wide range of things) – the crux of it (my understanding of it anyway), when you take it as “dishonest business drives honest business out of the market”, is that dishonest dealings that make a business more profitable, allow them to be more competitive than businesses that are honest, and in extreme cases this can allow them to takeover the market.
A good example is banking/finance without adequate regulations (which is Bill Black’s area of expertise – highly recommend his ‘The Best Way to Rob a Bank is to Own One’ book – not much new if you read his articles, but still very good), dishonest dealings (breaking regulations/laws, engaging in accounting control fraud to maximize short-term profits, in exchange for long-term insolvency) make many banking/financial institutions more competitive, to the point that they either buy up or drive out of business, the honest institutions that find it hard to compete, until you’re left with an industry disproportionately full of fraudulent institutions.
The direction you’re going here is interesting, so I’d be very interested to hear your take on all of that, especially checking out and applying your arguments to Bill Black’s writing/articles on the Greshams Dynamic.
The rise of fraud in the financial markets has nothing to do with what Akerlof is doing. He is arguing that people will form an average of the market value of assets and this average will be lowered due to the ‘lemons’.
In the financial markets the prices of assets did not fall due to fraud as Akerlof’s theory would predict. Rather they rose due to a wave of speculation and blindness.
I think that Tuckett’s approach is the only one that really captures these dynamics psychologically. I will write more on it in the next few weeks.
Ah okey, I’ll need to look into it more closely so – looking forward to those future articles.
Very interesting post.
Economists mostly seem to focus on single business transactions when discussing this type of thing. One aspect of business which economists rarely seem to discuss is ‘repeat business’. Repeat business is the sales you make to customers who have already purchased your products. If you sell a customer a good car then they are more likely to come back to you next time they want a car. They will also recommend you to their friends. The opposite can also occur. If you sell a customer a lemon they will not come back and they will tell their friends not to trust you. That’s how businesses establish good or bad reputations. In numerical terms, a satisfied customer base leads to a reduction in future marketing & sales costs per sale while a dissatisfied customer base leads to the opposite.
I’d say that the need for businesses to develop trust with their customers relies on the ability of customers to punish the absence of trust. Removal of repeat business is such a punishment. That’s why competition is so important. If I don’t trust my bank but my only alternatives are three or four other banks which I don’t trust either then I have lost my power to make the banks behave.
Trust is vital in pretty much everything we do. I believe that the universe started in a Big Bang. However, I would have no idea how to develop this idea for myself. What I am really saying is that I trust the physicists who developed this theory because physicists have a good track record in other areas which I do understand and because they present themselves in a very ethical way. In this case, trust overcomes a huge information asymmetry.
Once trust is lost (in a business or an institution or a political party or a profession) it can take a generation to repair that lack of trust. Look at sales of The Sun newspaper in Liverpool. 25 years on from Hillsborough, there is still no sign of trust being restored.
An important part of the ‘information asymmetry’ with banks and such, is that their dishonesty was not immediately visible to the public – so there was no way for public distrust to develop, to counteract their dishonesty – they were able to conceal/obfuscate the damage of their actions, and defer the consequences of that damage into the future, building them up to such unsustainable levels that it tanked economies and demanded bailouts.
By the time the damage came to light, it became too late for trust to matter at all, as by the time it became widely apparent just how dishonest/fraudulent many institutions had become, it was too late, because that industry was already disproportionately populated with dishonest/fraudulent business/dealings (and even captured disproportionate power over government).
In the real modern economy, does not asymmetry of information function more or less like asymmetry of soil fertility?
I.e., is it not just another form of Ricardian rent?
Let me spoil the party a little bit.
1) Akerlof doesn’t say that market with asymmetric information will always collapse, and if you think he does, you don’t understand the meaning and importance of his paper. He presents an example where market does collapse to make a point, not as a general statement.
2) Please, where do students spend “years” studying competitive supply and demand? Market failures are covered in introductory courses and textbooks right after competitive markets. Mankiw’s intro book, for example, does explicitly mention asymmetric information, in addition to the usual fare of monopolies, externalities and public goods. If you’ve received an undergraduate education which consisted of years of worshipping Marshallian scissors, you should ask for your money back – but somehow I doubt that’s the case.
3) I don’t think you understand what degenerative research program means. Your logic (or lack of logic) has a nice circular touch to it:
“Neoclassical econ should be abandoned in favor of X because it’s a degenerative research program.”
“Neoclassical econ is degenerative because faced with empirical failure, it adapts instead of being abandoned in favor of X”
That’s not how those terms work, at least according to Lakatos. That a theory adapts its auxiliary hypotheses is a normal way of doing science, as long as those adaptations are progressive – that is, they enhance predictive and explanatory power of the theory in a general way, beyond being ad-hoc fixes for the particular challenge which motivated their introduction. But to judge whether an adaption is progressive or not thus requires deeper evaluation of itself and its subsequent applications – and yes, that requires studying models, estimating regressions, reading (contemporary) papers and all that stuff you usually dismiss.
To be more specific, if Akerlof’s objective was to simply explain price differentials in used car market, and his proposed mechanism was not applicable anywhere else, one could perhaps call it degenerative. Anyone with even a hint of knowledge of post-war microeconomic theory knows otherwise – the asymmetric information idea led to huge number of diverse applications across both micro- and macroeconomics. Some of them are previewed right there in Akerlof’s own paper. Of course, you could still try to argue that the program is degenerative because all these applications are worthless… that is, if you had some idea they existed. But since you’ve repeatedly declared you’re not interested in studying details of mainstream theory, you can’t do even that.
I’ve always wondered what the big deal with “The Market for ‘Lemons'” paper was. This has clarified things considerably.
Out of interest, I finally got around to reading Hutchison’s “Hayek, Mises and the Methodological Contradictions of ‘Modern Austrian’ Economics”:
http://socialdemocracy21stcentury.blogspot.com.au/2014/08/hutchison-on-history-of-hayeks-views-on.html
All Akerlof’s model shows is that if you don’t assume perfect information but you do assume some vague notion of rationality the market will break down.
Does Akerlof’s model also assume that agents have rational expectations?
Yes. It’s a tedious exercise. I mean, the assumptions he is using to undermine the GE model are wholly arbitrary. If you assume imperfect information then why not assume non-rational behavior? It’s so dumb.