Friday, April 17, 2015

Solving market failures through information interventions

The standard approach in public economics

The standard approach of conventional public economics is to emphasise that the market economy generally works well, barring a group of market failures. The job of government is to try to address these market failures. The standard checklist of market failures is:
  1. Externalities -- e.g. a factory that pollutes.
  2. Asymmetric information -- e.g. safety in food or medicines.
  3. Market power -- e.g. firms that earn super-normal profit owing to weak competition, and
  4. Public goods -- e.g. law and order.

In the standard approach we weigh market failures against the problems of obtaining effective State intervention. The barrier is `public choice theory': the State is not benevolent. The citizen is the principal, the State is the agent, and there is a principal-agent problem. It is not easy to obtain performance when setting up real world arrangements that will sally forth, intervene in the working of the economy, and address market failures. This limits the class of situations where intervention may be appropriate. In a world with a perfect and benevolent State, we'd do a lot more in terms of going after market failures. In India, where State capacity is low, we are very selective and only do a few things.

The end of asymmetric information?

Alex Tabarrok and Tyler Cowen look at the proliferation of methods to create, store and transmit information, and say that there is an increased class of situations where asymmetric information has been conquered, thus reducing the extent of market failures.

Improved access to information also reduces the public choice problem. More information about the activities of politicians and bureaucrats is available to the citizen, which reduces the principal-agent problem.

It's a good article. I felt the case was a bit overstated. E.g. reputation measures on the Internet help people see more about you, and that's good in some settings (e.g. you know something about the Uber driver), but it's a small change in a vast gulf of lack of information. E.g. the authors say : Many public choice problems are really problems of asymmetric information. I don't agree. Yes, more information will help, but the principal-agent problem between citizen and State is vast and complicated. Merely monitoring some of the activities of civil servants better does not solve the public choice problem.
There is vast asymmetric information in the relationship between an employer and an employee in most complex work places, and nothing has changed which will make a dent to this. We can think of numerous other situations where the asymmetric information has not changed.

More or less government intervention?

The flavour of the article is that with less asymmetric information, you'd need less State. Yes, that is true, but it's also the case that with less of a public choice problem, you could use more State. In thinking about public policy, we are constantly watching the market failure that's worth addressing versus our ability to construct a State apparatus that would actually deliver the goods in trying to address this market failure. The new age of improved information cuts both ways: it reduces the places where we might want a State intervention, and it increases the class of places where we could pull off a successful State intervention.

A new kind of State intervention

In this new age of easy capture, storage and transmission of information, I have felt that there is a new kind of State intervention: one which rearranges the information set. The State can use its coercive power to force certain kind of information to be captured or released or transmitted. This is a beautiful intervention that directly addresses the root cause of the market failure, the lack of information.

As an example, in the old analysis of insurance, there were some good drivers and some bad drivers, but the insurance company did not know who fell into what category. There was adverse selection (bad drivers were more likely to sign up for insurance) which led to high prices of insurance and many good drivers got insurance which was not actuarially fair.

This is the standard description of the market failure, in the textbooks. We can now think of a new kind of State intervention: The government forces cars to be equipped with devices that measure how the person drives. This is an intervention that directly stabs at the asymmetric information.

Here is another example which nicely illustrates an information intervention. We have been working in the field of regulation of warehouses. There is an asymmetric information problem: I submit my goods for storage at the warehouse, but I don't know how much care will be exercised by the warehouseman. There are old style interventions which reduce this information asymmetry.
In some situations, we can directly attack the information asymmetry. As an example, consider frozen food. When you deposit 1000 kilograms of cheese into a cold storage, you worry that the warehouseman will not maintain the temperature at precisely 4 degrees. But now there are low cost devices that will measure the temperature every minute and thus tell you what your cheese experienced.

Now the customer and the warehouseman can enter into a private contract where the temperature of the cheese is monitored, and a set of payoffs calculated based on the extent to which the temperature of the cheese strays above 4 degrees. Good warehousemen would think: Why don't I release this information, so that prospective customers would trust me? The trouble is: this data could be tampered with, or data could be selectively released.

This suggests the design of a government intervention: The government could establish an inspection mechanism which ensures truthful release of comprehensive data about all transactions by the warehouse. This is a combination of the new age of devices (the data logger) plus a dose of State intervention (to ensure truthful and complete data release). We could also envision a valuable State intervention that standardises the XML files which are put out by all warehousemen, which would reduce the cost of processing this data for customers.

This is an example of what I call an `information intervention', which rearranges the structure of information, and thus combats the market failure that's rooted in asymmetric information.


  1. In the "Please note" on this blog it should also be clarified that users who wish to comment should first sign-in and only then type in their comments or else comments will be irretrievably lost. I have had this happen to me more than once here since this is a somewhat unusual feature of the software that drives this particular blog.

    1. Thanks. We are trying to figure out what's going wrong.

  2. This is an excellent discussion. In healthcare a similar information intervention could be the implantation of a health tracker in all of us from birth so that we become more fully informed about our "true but unobserved" health status and therefore better participants in healthcare markets.

    However, isn't there a more a permanent market failure which comes from the differences between what the smartest people/institutions can do in a population (moore's law for example) and the analytical and cognitive capabilities of the general population, and from the presence of hyperbolic discounting imposed upon all of us by our frontal cortex which is presumably a natural evolutionary response to the presence of mortality (the charvaka philosophy is an example of hyperbolic preferences in action). To correct the effects of these failures wouldn't the State need to act in a paternalistic manner in a more permanent way by, for example, imposing tobacco taxes with unfiltered tobacco being taxed at a higher rate; taxing saturated fats and sugars in foods at a higher rate and perhaps allowing for "negative taxation" on healthier foods; imposing "suitability" requirements on financial services providers; and allowing healthcare providers (and consumers) to operate only within rigidly defined health systems designs which may, for example, require primary care providers to emerge as mandatory gatekeepers.

    1. I think a useful way to decompose the overall public economics problem is: Rational humans versus behavioural economics.

      In a rational world, the humans make sound choices. Consumers are sensible users of information. Politicians and officials are rational and try to maximise their own self interest. That's the standard world of the microeconomics textbooks.

      It's hard to make government work under these conditions. On top of that, when you come to Mark 2, where humans suffer from cognitive problems, it's a whole new problem and it's much harder.

      For us in India, as a thumb rule, I think we should put our prime focus on the first (simple) case.

      As an example, monetary policy is fairly straightforward. Setup an inflation target, setup a sound MPC, and the first order problem is solved. India is not there. We suffer from inflation going between 2% and 16%. These first order reforms will get us to inflation between 2% and 6%. This will be huge progress.

      These "trivial and first order reforms" involve confronting RBI's biases in favour of low accountability and protecting incompetence, and nothing more. E.g. RBI campaigns against setting up a sensible MPC. These are easy problems to solve.

      Into this happy world, you introduce the cognitive problems of households. An example is: IFC Consumer Protection. In comparison, IFC consumer protection is truly hard. We have to chip away, of course, but the problems are difficult and building State capacity will be hard work.

      The scientific foundations required for doing IFC consumer protection are also quite weak. A great wave of research papers is required and so far there is only a tiny literature.

      It will be truly hard to build capable bureaucracies which will operate IFC consumer protection.


Please note: Comments are moderated. Only civilised conversation is permitted on this blog. Criticising me is perfectly okay; uncivilised language is not. I delete any comment which is spam, has personal attacks against anyone, or uses foul language. I delete any comment which does not contribute to the intellectual discussion about the blog article in question.

Please note: LaTeX mathematics works. This means that if you want to say $10 you have to say \$10.