The Economics of Networks. (Reviews/Comptes Rendus).
The authors of the papers in this book describe networks as economic systems in which agents interact directly with each other rather than only indirectly through a decentralised price system. Agents communicate with each other, exchange information, infer things from what others do, imitate each other, and get payoffs that depend directly on what others in some reference group do. All these make it dangerous to infer macro qualities from some average or representative agent, as is common in other parts of economics.
Most economists know by now that in models of networks there can be all sorts of interesting phenomena, and the authors discuss all of these: network externalities or "adoption externalities," leading to increasing returns; inefficient standardisation of technology; multiple equilibria and equilibria that depend on the exact sequence of choices made by actors (in sharp distinction from the Walrasian price system model); niches or enclaves of technique; phase transitions with sharp discontinuities and sensitivity to threshold values of parameters; bubbles; dynamics subject to a few errors early in an evolutionary process; information cascades, in which it is optimal for people to copy others rather than act on their own private information (in Chapter 13, Marc Willinger and Anthony Ziegelmeyer report on cascades in some lab experiments). The authors appropriately pay homage to Thomas Schelling (for example, Schelling 1971). One of the nuggets in the book is an alternative phrase for network externalities: "a premium for the alignment of choices" (Jean-Michel Dalle, Chapter 4:129, quoting Dalle and Foray 1995).
A critical distinction is between "global" interaction and "local" interaction. In the global case, an agent can interact with any other agent in the economy being analysed, in the local case he or she has a location and interacts only with others "near" him or her and in the same "neighbourhood." Many of the authors focus on differences between the global and the local case, and on the implications of different assumptions about the size and structure of neighbourhoods.
The book has 16 essays. Almost all the authors are French, and their efforts were coordinated by BETA (Bureau d'Economie Theorique et Appliquee) at the Universite Louis Pasteur of Strasbourg. The essays vary considerably in intent and style; some are mainly summaries of the literature and general issues and do not present new results, while others present and work through original models in detail. In the latter case, the models are sometimes too complex to be solved analytically and so the authors report numerical simulations. A very common application is the diffusion and survival of competing technologies, following in the path-dependency and lock-in tradition of Brian Arthur and Paul David, but one paper is on the diffusion and survival of cooperation in a prisoner's dilemma game (Vanessa Oltra and Eric Schenk, Chapter 8). Game theory is pervasive, and a reader not already well versed in it will find much of the book tough going.
The authors concerned with local interactions specify some rudimentary spatial structure, making the character of both interaction (e.g., imitation) and network externalities depend on the distance between the agents. For example,. Nicolas Jonard, Patrick Lierena and Babak Mehmanpazir (Chapter 5) put their agents, who are adopting a technology, on a circle; several others put agents on a torus, as for example in Jonard and Marat Yildzizoglu's model of NelsonWinter firms (Chapter 7). However, authors take pains to point out briefly that the spatial structure need not be geographic, but rather due to social proximity (e.g., due to friendships or social ladders; see Gisele Umbhauer's essay (Chapter 6)). Distance has an effect on the various processes of diffusion and survival the authors are analysing, a common result being that small distances preserve diversity while larger ones tend to produce standardisation. On the other hand, I do not remember a single model in the book that specifies any characteristic o ther than distance that affects costs of interaction, as would be true if the links are roads or pipes or wires.
Jonard and Yildzizoglu's firms have a more complicated production situation than the firms in the other models in the book: the firms are imperfect competitors, and they must exit if they do not make adequate profits; they make technology decisions by R&D investment (both imitative and innovative) and also make physical capital investments; and of course they are affected by local externalities of technology use. The authors find that "very localized network externalities mainly sustain diversity by softening the [competitive] selection in the early history of the industry" (p. 190) and "the presence of very localized network externalities is not necessarily harmful to technological diversity" (p. 196).
Many models incorporate learning; some also incorporate mutation, because in the real world agents die and are replaced by new ones who have no history and so choose randomly, or agents deliberately experiment, or they cannot always calculate the best action (due to bounded rationality or lack of information) (Umbhauer, Chapter 6:170-1).
A good feature of the book is that the authors are humble, in the sense of making the reader aware of shortcomings in their own models and others in the literature. Dalle, for example, has a stimulating (though quite ponderous) critique of the limited "rationality" exhibited by the actors in many standard models of network interaction (Chapter 4). A bad feature of the book is the very poor copyediting; it is not a book Springer can be proud of. There are scores of typos, ambiguities, and poor translations from French to English. There is no index, which is unfortunate since many authors discuss the same topic but in slightly different ways. In one essay, there are neither line spaces nor indents to signal paragraph breaks, there are no equations to help break up the mass of text, and one paragraph is over a page long. Many of the howlers of course will not derail, though they may amuse, experienced scholars, but in fairness I must note that there are also errors that make things hard for the very reader I as sume the editors hoped to appeal to.
I can recommend the book as a whole only to experienced scholars who already know a fair bit about the subjects being discussed. It is intended to be an advanced book; the models are simplified but the mathematical notation is complex, the authors often are terse in summarising prior literature and traditions, and quite a few papers build on previous work of the authors that they do not explain in detail. But for experienced scholars some of the essays are worthwhile, and the collection as a whole gives a good sense of the sorts of models that are out there. Every essay except the introduction has a valuable list of references.
Dalle, J.-M. and D. Foray. 1995. "Des fourmis et des hommes: modele stochastiques d'interactions et rationalite individuelle active". Cahiers d'Economie et de Sociologie Rurale, 37: 70-92.
Schelling, Thomas C. 1971. "Dynamic Models of Segregation". Journal of Mathematical Sociology, 1:143-86.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Interaction and Behaviours|
|Publication:||Canadian Journal of Regional Science|
|Article Type:||Book Review|
|Date:||Jun 22, 2001|
|Previous Article:||Amassing Power. J.B. Duke and the Saguenay River, 1897-1927. (Reviews/Comptes Rendus).|
|Next Article:||Obligation and Opportunity: Single Maritime Women in Boston, 1870-1930 & A Sense of Their Duty: Middle-Class Formation in Victorian Ontario Towns....|