On Stigler's Economics of Information

<a href="http://home.uchicago.edu/~vlima/courses/econ200/spring01/stigler.pdf">The Economics of Information is a highly influential academic article written by the economist George Stigler. It was published by the Journal of Political Economy in 1961. This article is my review of his article.

To give away the ending: I am a huge fan of this article, but I do have occasional questions or disagreements.

• Stigler notes the irony that academics make a profession of knowledge and yet economics occupies \"a slum dwelling in the town\" of formal economic literature.
• Stigler argues that it is useful to analyze some questions in economics using a concept he calls search. This particular paper analyzes the emergence of market price as the result of a search process.
• What is search? For Stigler, search is the process of a buyer or seller attempting to find the most favorable price. In plain English we might call this ‘bargain hunting’.
• Stigler notes the empirical fact of price dispersion, even for homogenous goods. That is, even identical goods are supplied and demanded at a variety of prices.
• Stigler argues ‘price dispersion is a manifestation, and indeed a measure, of ignorance in the market’.
• The idea here is that a better informed market will have less price dispersion. I’m not sure that always holds but it seems reasonable as a general rule.
• Stigler says ‘the frequency distributions of asking and offering prices have not been studied sufficiently to support any hypothesis as to their nature.’
• But then he goes on to theorize that price dispersion of asks (that is, the price the producer requests) is probably skewed to the right as sellers will have a minimum price but not maximum.
• I'm not sure Stigler thought the last point through very well because there are minimum and maximum prices and they are equal for buyers and sellers. That is, the minimum price is 0 and the maximum price is all the money in the economy.
• Another issue I have with the Stigler article is that he supposes consumers will search for the minimum price and producers will search for the maximum sale price. In doing so he ignores two key facts about reality.
  • First, consumers often engage in charity. In pay what you want transactions, consumers are often allowed to obtain a good for free, but they will pay anyway.
  • Second, many producers and consumers are threshold-searchers. Instead of searching until marginal benefit equals marginal cost, they will search until total benefits equal target-value benefits.
• Stigler's theory summarily says that the process of search has expected costs and benefits. He predictably argues that individuals should search until the net marginal benefit of search is 0.
• After laying out the obvious he gets into some cool stuff but only after saying one thing I disagree with.
• He says \"...it is certain that increased search will yield diminishing returns...\" I strongly disagree. Research can be analyzed as a Stiglar search process, and the result is that such a search sometimes produces constant or increasing returns to scale.
• Another issue I have with the article, and mainstream economics at large, is the producer-consumer paradigm. This view ignores the possibility of a consumer who searches outside the market and produces for himself. That person would be like a subsistence farmer, an entrepreneur, or a programmer who develops a new program because the function can't be bought anywhere, or it could be self-produced for less than the market price.
• Some of the cool stuff Stigler goes on to discuss includes the fact that identity and reputation of buyers and sellers are key signals of risk affecting price in transactions. He even gets into theory of advertising by identifying advertising and marketing as the method producers use for search. Today we might include digital analytics in this camp. He touches on all of this briefly but doesn’t formalize it much.
• He gives special attention to the distribution of prices, which can come in a variety of shapes. The shape of the distribution, he claims, depends on the returns to scale for the agents in the market. I definitely agree with him there. One thing I didn’t like is that. He did not offer an example of an equilibrium model which is robust to distributional variance.
• He discusses determinants of search, which is the cost and benefit. He notes some sources of search costs, such as the costs of a firm figuring out competitor’s prices and also advertising costs.
• One thing I don’t like is that he has this huge discussion on information without once discussing the quality of information, the variety of significance of heterogeneous information, or the truth-value of information.
• Clearly true information allows producers and consumers to make better decisions compared to false information.
• Clearly some information is more important than other information. If I sell cars, should I advertise that the McDonald's down the street is giving away free breakfast all day? The irrelevance of such information makes it less effective and less valuable advertising.
• In other words he basically comes to the conclusion ‘advertising matters, so firms should spend money on advertising,’ but he says next to nothing about what constitutes good advertising. Is it in the interest of firms to produce deceitful or truthful messages?
• Basically, advertising has heterogeneous return based on certain factors, and this is ignored by Stigler.
• Similarly, search has a heterogeneous return for consumers. Stigler doesn’t discern well between effective and ineffective search. He basically just says, ‘So ya. People should do the search.’ There is little practical advice regarding efficient ways to search.
• The last thing I would point out is that Stigler discounts the differences between real values and perceived values. He doesn’t get into a discussion about how advertising can stimulate malconsumption, for example, by misleading individuals.
• In conclusion this paper contains some awesome theory and some really good formalization of an ambiguous search cost term. This is the sort of paper I would love to see come out today because it at once makes a great point and also opens the floodgates for new questions. I think if we fast forward to today, many of the questions Stigler stimulated have already been answered. The literature answers what constitutes good and bad methods of advertising and bargain hunting today.
• I also think that, at least so far, I haven’t seen any mainstream models which really take into account everything Stigler discussed. Part of the reason may be that such a model would be immensely complex. But we have cool programming languages these days, so it would well be worth constructing and I think it would yield results which are more accurate than the typical RCK model or whatever else we’ve discussed so far.