All but one: How pioneers of linear economics overlooked Perron-Frobenius mathematics Wilfried PARYS Paper prepared for the Conference “The Pioneers of Linear Models of Production” at the University of Paris Ouest, Nanterre, 17-18 January 2013 Address for correspondence. Wilfried Parys, Department of Economics, University of Antwerp, Prinsstraat 13, 2000 Antwerp, Belgium;
[email protected] 1. Introduction Recently MIT Press published the final two volumes, numbers 6 and 7, of The Collected Scientific Papers of Paul Anthony Samuelson. Because he wrote on many and widely different topics, Samuelson has often been considered the last generalist in economics, but it is remarkable how large a proportion of the final two volumes are devoted to linear economics.1 Besides Samuelson, at least nine other 20th century Nobel laureates in economics (Wassily Leontief, Ragnar Frisch, Herbert Simon, Tjalling Koopmans, Kenneth Arrow, Robert Solow, Gérard Debreu, John Hicks, Richard Stone) published interesting work on similar linear models.2 Because the inputs in processes of production are either positive or zero, the theory of positive or nonnegative matrices plays a crucial role in many modern treatments of linear economics. The title of my paper refers to “Perron-Frobenius mathematics” in honour of the fundamental papers by Oskar Perron (1907a, 1907b) and Georg Frobenius (1908, 1909, 1912) on positive and on nonnegative matrices.3 Today the Perron-Frobenius mathematics of nonnegative matrices enjoys wide applications, the most sensational perhaps being its implicit use by millions of internet surfers, who routinely activate Google’s PageRank algorithm every day (Langville & Meyer, 2006). Such a remarkable worldwide application in an electronic search engine is a far cry from the original theoretical articles, written more than a century ago, by Perron and Frobenius.