Letter from Kenneth J. Arrow, January 17, 1998

To Michael Kinsley, Slate Magazine:

Paul Krugman's attack on Brian Arthur in the January 14 issue of your journal requires a correction of its misrepresentations of fact. Arthur is a reputable and significant scholar whose work is indeed having influence in the field of industrial organization and in particular public policy towards anti-trust policy in high-tech industries. Krugman admits that he wrote the article because he was "just pissed off," not a very good state for a judicious statement of facts, as his column shows.

His theme is stated in his first paragraph: "Cassidy's article [in the New Yorker of Jan 12] tells the story of how Stanford Professor Brian Arthur came up with the idea of increasing returns." Cassidy however said nothing of the sort. The concept of increasing returns is indeed very old, and Cassidy at no point attributed that idea to Arthur. Indeed, the phrase, "increasing returns," appears just once in Cassidy's article and then merely to say that Arthur had used the term while others refer to network externalities. Further, Arthur has never made any such preposterous claim at any other time. On the contrary, his papers have fully cited the history of the field and made references to the previous papers, including those of Paul Krugman (see Arthur's papers collected in the volume, Increasing Returns and Path Dependence in the Economy, University of Michigan Press, 1994, and especially his Preface and my Foreword for longer comments on Arthur's work in historic perspective). Hence, Krugman's whole attack is directed at a statement made neither by Arthur nor by Cassidy. Krugman has not read Cassidy's piece with any care nor has he bothered to review what Arthur has in fact said.

What Cassidy in fact did in his article was to trace a line of influence between one of Arthur's early articles and the current claims of the Department of Justice against Microsoft. It appears that Cassidy based his article on several interviews, not just one.

The point that Arthur has emphasized and which is influential in the current debates about anti-trust policy is the dynamic implication of increasing returns. It is the concept of path-dependence, that small events, whether random or the result of corporate strategic choice, may have large consequences because of increasing returns of various kinds. Initial small advantages become magnified, for example, by creating a large installed base and direct the future, possibly in an inefficient direction. Techniques of production may be locked in at an early stage. Similar considerations apply to regional development and learning.

Sincerely yours,

Kenneth J. Arrow, Nobel Laureate, and Joan Kenney Professor of Economics Emeritus, Stanford University