Customer Reviews

Sunk Costs and Market Structure: Price Competition, Advertising, and the Evolution of Concentration

5 star:		(1)
4 star:		(2)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

I studied this book in my first year of graduate school in economics in a course on Industrial Organization. I highly recommend this book. It is NOT a text book, it does not review the state of the art in I.O., it is Sutton's theory about market structure and concentration as shaped by EXOgenous barriers to entry.

I liked the book because it tries to explain market structure using OBSERVABLE variables, because it tests the theory both via case study and via econometric methods and because it is full of innovative ideas. I want to mention in particular the way in which Sutton tackles the problem of multiplicity of equilibria, what he calls "the bound approach"

A few months ago Sutton published a companion book called Technology and Market Structure in which he carries over the same line of research to endogenous barriers to entry.

John Sutton tells us that there's a tension in the study of why industries are structured the way they are. If you want to study, say, why the microprocessor industry is concentrated (closer to monopoly) whereas the cement industry is not, you can come up with models that contain lots of detail about individual industries and are therefore not of much general use; or you can put forth very vague assumptions that apply across the general run of companies. _Sunk Costs and Market Structure_ is an attempt to strike a good compromise between these two extremes.

Sutton's guiding observation is that if an industry is monopolistic in one country, it tends to be monopolistic in another. This is not obviously the way things had to turn out. It holds across a broad swath of industries, as well, which means it's an empirical regularity that any interesting industrial-structure model has to explain.

Sutton goes about explaining it by dividing industries into two categories and deriving substantially different conclusions about each:

* Industries in which advertising is not important. These industries behave as you'd expect: as market size increases, concentration decreases, regardless of the size of fixed costs. Take microprocessors, for instance. It costs billions of dollars to establish a new fabrication facility to compete on the proper scale with Intel. This is an imposing fixed cost, and will keep out many potential entrants. But imagine now that the market size grows without bound -- that the microprocessor market reaches trillions of dollars. Then that fixed cost comes to seem less and less significant, and turns into a smaller and smaller barrier to entry. As market size grows, then, we'd expect more entrants. More specifically, Sutton shows that as market size increases, the fraction of market share held by the largest company in that industry tends to zero.

There's a bit of detail that I need to explain here. (Indeed, there are many bits of detail in _Sunk Costs and Market Structure_; I can only explain a small fraction of them here. You should really read the book.) We need to explain what we mean by "compete on the proper scale with Intel." If you start a small factory to produce a few thousand microprocessors, you'll need to recoup your fixed investment in that factory, so the average cost per microprocessor will equal something like the total cost of the factory divided by the number of microprocessors you can crank out over the factory's lifetime. Intel, being much larger than you, builds much larger factories. In particular, they may build a factory several thousand times as large, whose per-processor costs are much lower than yours; we say that Intel exploits "increasing returns to scale" (i.e., "there are advantages to being big"). Their processors will therefore sell for far less than yours, and you will be driven out of the market.

There is, then, a "minimum efficient scale" (m.e.s.) in a given industry -- a factory size below which you can't produce products at a price that anyone would buy. There are various ways to define minimum efficient scale, all of which Sutton discusses fairly. We can ignore here the specific measures of m.e.s. that he settles on, but suffice it to say that for industries in which advertising is not important, the fixed cost of building a m.e.s. factory is what concerns him.
* Advertising-intensive industries. Sutton describes advertising as a cost that increases the perceived quality of the product, and therefore increases the amount which customers are willing to pay for it. Under this description of advertising, Sutton shows that advertising-intensive industries must always be concentrated above a fixed threshold. That threshold is universal in the sense that, knowing nothing else about a given industry other than that industry's responsiveness to increased advertising, you can assert that its concentration level will never fall below a fixed level.

What's most remarkable about this result is that it holds across a wide variety of models. Sutton explores a large set of such models from the industrial-concentration literature; the man clearly knows the field inside and out. He abstracts away from these models with two simple axioms that describe customer response to advertising.

Sutton spends the rest of the book very carefully analyzing a wide range of industries for conformance with his theory. The industries studied range from ready-to-eat cereals to canned goods to sugar to salt to beer. Each has varying responses to advertising and varying fixed costs (for instance, higher or lower levels of R&D). Within specific industries, even, these quantities often vary: sales of frozen foods sold directly to consumers respond to advertising, whereas those sold to restaurants and caterers, for instance, behave more like commodity goods sold on price. The industry profiles forming the latter 2/3 of the book, based on interviews with executives at many companies and on widely available industry statistics, are a fascinating read on their own, and can be read apart from the heavily mathematical first third of the book.

The theory stands up reasonably well to the evidence, particularly given the paucity of data on many industries. The data are cleanest for the frozen-foods industry, whose evolution -- overtaking the canned-goods industry half a century ago -- is well-documented. Maybe more the point, Sutton observes that no more-detailed theory could apply to such a wide run of industries.

This book should be understood in the context of Sutton's broader project, as laid out for a broader audience in _Marshall's Tendencies_. Economics has tried for at least a few decades to model The One True Equilibrium for any number of processes: how bargaining works, how industries are structured, and how cities are formed, among many others. Indeed, Fujita, Krugman, and Venables repeatedly assert in _The Spatial Economy_ that their idealizations are meant to make their models tractable -- which is to say, meant to produce a model which will lead to a few concrete equilibria. Sutton points the obvious: too many assumptions are necessary to make this exercise work out. The consequence of assuming so much is that the models fall over when touched with a feather. Instead of trying to nail down equilibria, set up deliberately broad, but highly believable axioms, and see where that takes you. Good, broad axioms won't take you all the way to specific equilibria, but they'll lead to a *class of models* that's still useful and narrowly circumscribed -- and, more to the point, is likely to hold up under scrutiny in a way that more-precise models do not.

Seen as a demonstration of *method* rather than as a cookbook, _Sunk Costs and Market Structure_ really ought to be viewed as a revolution in economic practice. Combine this with its scrupulous fairness toward competing models, its analytical rigour, and its abundance of data, and you end up with an economic must-read.

While the material is a little dated (as is my recollection of the book), but this book provides an excellent overview of the empirical evidence regarding advertising from an industrial organization perspective. A theoretical context is provided and related to the empirical literature. This is a useful introduction for someone looking to do research on advertising economics.