How not to write about asymmetric information. (updated)

Recently I read a paper on why business schools in some universities, including those in New Zealand, seek to gain accreditation from organisations such as the Association of MBAs, Association for the Advancement of Collegiate Schools of Business (AACSB) International and the European Foundation for Management Development. The paper examined at number of topics but the section that interested me most was on "Information Asymmetry" and its relevance to accreditation.

The section opens with the statement,

The problems associated with asymmetric information were first analysed by economist Kenneth J. Arrow (1963) ...

But then what was Adam Smith doing back in 1776 when he wrote about the lack of proper incentives inherent in slavery,

the work done by slaves, though it appears to cost only their maintenance, is in the end the dearest of any. A person who can acquire no property, can have no other interest but to eat as much, and to labour as little as possible.

Or in his discussion of the incentives facing directors of joint stock companies,

The directors of such companies, however, being the managers rather of other people's money than of their own, it cannot well be expected that they should watch over it with the same anxious vigilance with which the partners in a private copartnery frequently watch over their own. Like the stewards of a rich man, they are apt to consider attention to small matters as not for their master's honour, and very easily give themselves a dispensation from having it. Negligence and profusion, therefore, must always prevail, more or less, in the management of the affairs of such a company.

Also what are we to make of Hayek (1937) and (1945). If these papers are not about asymmetric information then what are they about? Berle and Means (1932) examined the problems arising from the separation of ownership and management. Barnard (1938) was an early attempt to develop a general theory of incentives in management. All before Arrow.

Then we are told that

Sometimes referred to as information asymmetry, Arrow's theory describes transactions where one party (usually the seller) has more information about the product being sold than the other party.

Of course today we know the issue is much wider than that. Note that the whole moral hazard literature seems to be excluded under the above statement.

Then we find out that

Information asymmetry leads to the typical "agency" problem. An agency relationship occurs when one individual (the principal) depends on the action of another (the agent). Difficulties arise when the agent has more information than the principal, or the principal cannot perfectly and costlessly monitor the agent’s action and information (Ro, 1988).

These last two sentences don't sit well with each other. The first describes a situation of moral hazard with hidden action while the second could apply to moral hazard with hidden action, moral hazard with hidden information or adverse selection. In fact one of the problems with this whole section is that the distinction between these three different problems resulting from asymmetric information is never made. And at different times in the discussion each or any of the three could be what is meant. Also, it should be noted that even if the principal can perfectly and costlessly monitor the agent's action and information, but this can not be verified to an outside third party, eg the courts, then problems of incompleteness of contract can occur. An issue not considered in this section. With regard to the notion that

Difficulties arise when the agent has more information than the principal...

it should be noted that while this is true, it is also true that difficulties arise when the principal has more information than the agent. So why is it not relevant here?

Following on we learn that,

Ro further contends that "information is a primary source of transaction costs."

This also is true but its relevance to the discussion is not made clear. Also problems with transaction costs can give rise to incomplete contracts rather than the comprehensive contracts discussed in this section. The example given at this point is

... sales representatives are employed, and advertising undertaken, so that businesses can relay information about the quality of goods they are selling. The dissemination of that information is a cost to the business. The information conveyed, however, tells the prospective purchaser (principal) that the company (agent) is committing significant resources to the product and has confidence of its quality.

But are these costs transaction costs or just costs of production? As Allen (1999) makes clear there are two definitions of transaction costs: Transaction Costs 1: the costs establishing and maintaining property rights; Transaction Costs 2: the costs resulting from the transfer of property rights. It is not clear how the example falls under either definition.

Next we learn that in the tertiary education sector there are two possible (adverse selection) problems that can rise. First there is the relationship between the prospective student (the principal) and the school (the agent) and the second between the institution and prospective employee, and in the second case, the information asymmetry can work in both directions.

The institution generally has incomplete information about the merits of the applicant it is about to employ and, likewise, the applicant may have incomplete information about the merits of the institution.

So if information is incomplete, but symmetrically so, where is the problem? Think of the Akerlof model where neither the seller nor the buyer knows the true state of the car, would not all cars sell at the average price of good and bad cars? The problem here must be that each party is unsure about a different things, the university about some characteristic of the employee and the employee about some characteristic of the university. But if this is correct signalling via accreditation will, at best, help with one of these problems, the employee's uncertainty about the university. The other issue still remains.

Later we are told

they [the university] are not in a position to predict the motivation or subject-specific ability of the student and, conversely, the student has no way of knowing how well the lecturer will be able to teach.

What is the problem here? Is it one of adverse section, the student just can't pass the course for some externally determined reason (ability), or is it moral hazard with hidden action, in that they do little work (motivation)? Also with regard to the teaching of the lecturer, is the problem again adverse selection, they just can't teach, or moral hazard, they don't want to teach? The solution will depend on the question.

Next we see that

Asymmetric information creates incentives for the party with more information to cheat the party with less information. As a result, a number of market structures have developed, which enable markets with asymmetric information to function. Those structures include: guarantees or warranties (Arrow, 1963, and Akerlof, 1970); brand name goods (Akerlof, 1970); and third party authentication (Arrow, 1963).

The market responses noted are responses not to asymmetric information in general but are responses to adverse selection in particular. They do nothing to mitigate the problems of moral hazard.

We are then told that

Licensing (Akerlof, 1970) and accreditation are examples of third party authentication which provide the market some measure of assurance of the quality of the product and the institution.

and that

Investment in accreditation, therefore, signals to potential students that the institution is committed to providing a high quality education and is committed to remaining in the market long term. This is a clear indication of the institution’s desire to enhance its reputation in the marketplace.

How does accreditation do this? Third party accreditation works when the uninformed party believes the assessment of third party. Why would students believe some accreditation party they have never heard of, know nothing about, who uses a process they know nothing about and they have no ability to verify the results of? Given that it is not realistic for students to be perfectly informed about the university, why is it any more realistic for them to be informed about the accreditation process? All that is happening is that one asymmetric information problem, between the students and the university, is being replaced by another between the students and the accreditation body.

Also, Is this the most efficient way to obtain the desired results? How do non-accredited universities maintain their reputation for quality? Why have other methods to deal with adverse selection not been examined in this section?

It is noted next that

Agency relationships exist between students and the university, whereby the student becomes the agent and depends on the lecturer (or the university), as the principal, to provide a good quality education.

Again, what is the problem here, adverse selection, moral hazard with hidden action or moral hazard with hidden information? The answer to the problem will depend on the nature of the problem, so it must be clear as to what the problem is.

Following this it is pointed out that,

The university therefore uses a variety of methods such as advertising and third party authentication, in the form of accreditation, to signal its quality and improve the flow of information.

But to what questions are these the answer? Its not clear how third party authentication will help deal with moral hazard and it may not even deal with adverse selection. What empirical evidence is there that such accreditation is in fact an efficient way of mitigating adverse selection?

Later, we are told

Joining associations such as AMBA, EFMD and AACSB, and achieving accreditation, enables institutions to signal their quality and commitment to the market and thus enhance their legitimacy.

If so why isn't the University of Chicago or Harvard or Stanford or the London Business School so accredited?

More general issues raised by this section are, first, much of the discussion is in terms of a world of uncertainty but the modelling is in terms of risk. How is this justified? The issues discussed in this section are issues where genuine uncertainty is relevant, so why are they modelled in terms of risk? Secondly, if the point of accreditation as a signal is to make the university "stand out" from the rest, but the rest also have accreditation, where is the value in the signal? In the section of the paper under discussion it seems that a signal is useful only in so far as it can separate agents, which can not happen if all agents use the same signal. That is, only separating equilibria are considered. What are we to make of a pooling equilibrium, which is where in New Zealand we look to be heading, in terms of the discussion in this section? Thirdly, in the Spence (1973) model, mentioned in this section, it is important that the costs of the signal differ across different types, but is this a sensible assumption in the case of accreditation of universities? In Spence model the signal works because the costs involved in signalling are higher, in some sense, for "low quality" types than for "high quality" types. This is the single crossing property. But are there large differences in the "cost" to each university in obtaining accreditation? What does low and high quality mean in this context, what differentiates "types" here? Fourth, is the whole problem looked at the right one? Following Arrow (1973), what if the value of education to the student is as a signal, and it does not increase your productivity. That signal being that you can get into the university, thereby signalling that you are smart, and what goes on during the university education is less valuable, then what use is accreditation? Would a better signal not be a very tough entrance exam? Gaining entrance would then act as a strong signal as to how smart the student is. Of course admission isn't the only signal that university study provides. A second signal is via grades and graduation. The point here is that the signal works even when the student does not learn anything useful with regard to their ultimate occupation. The key to signalling is that the signalling activity is easier for the desirable type to do well at than it is for the less desirable type. For example, in the selection of future managers, there is no reason that the activity cannot be the leaning of Latin or the study of philosophy, provided that those who will make the best managers find these subjects easier than those who will be bad managers. If fact as R. Preston McAfee has explained

Making the study useful, in fact, can be positively harmful, if the most desirable type finds the subject so tedious that they do not perform well. (McAfee 2002: 330).

Thus in so far as this is true, its not clear what benefits flow from accreditation to students via education as a signal.

• Akerlof, G. A. (1970). "The Market for "Lemons": Quality Uncertainty and the Market Mechanism". Quarterly Journal of Economics, 84(3), 488-500.


• Allen, D. (1999). "Transaction Costs". Encyclopedia of Law and Economics.


• Arrow, Kenneth J. (1963). "Uncertainty and the Welfare Economics of Medical Care". The American Economic Review. Vol. 53, No. 5. (Dec.), pp. 941-973.


• Arrow, Kenneth J. (1973). "Higher Education as a Filter". Journal of Public Economics. Vol. 2, No. 3. (july), pp. 193-216.


• Barnard, C. (1938). "Functions of the Executive". Cambridge: Harvard University Press.


• Berle, Adolf Augustus, Jr. and Gardiner Means (1932). "The Modern Corporation and Private Property". New York: Macmillan.


• Hayek, F. A. (1937). "Economics and Knowledge". Economica. Vol. 4. (Feb.) pp. 33-54.


• Hayek, F. A. (1945). "The Use of Knowledge in Society". The American Economic Review. Vol. 35, No. 4. (Sept): 519-530.


• McAfee, R. Prestion (2002). "Competitive Solutions: The Strategist's Toolkit". Princeton: Princeton University Press.


• Ro, S. (1988). "Market Organization and Product Quality under Asymmetric Information: A Comparison of Monopoly and Competitive Market". Thesis. Texas A&M University.


• Spence, Michael (1973). "Job Market Signaling". The Quarterly Journal of Economics. Vol. 87, No. 3. (Aug.): 355-374.
  

Update: For discussions of the signalling model of education see Tyler Cowen and Bryan Caplan.