_____________ _________ Luiz Carlos Bresser-Pereira is emeritus professor of Getulio Vargas Foundation. lcbresser@uol.com.br www.bresserpereira.org.br This paper was written based on the course of Scientific Methodology for Economists that I teach since 1989. My special thanks to my students in this course, and particularly to Ramón Garcia Fernandez and José Marcio Rego, who divide with me the responsibility for it. I am grateful for their comments, and for those of Adam Przeworski, Victoria Chick Alain Herscovici, Gilberto Tadeu Lima, Marcos Ribeiro Ferrari, Paulo Gala, Robert Nicol, Solange Marin and Yoshiaki Nakano. ECONOMICS' TWO METHODS Luiz Carlos Bresser-Pereira Paper presented at the European Association for Evolutionary Political Economy XVth Annual Conference, Maastricht, November 7- 10, 2003. Revised in November 2005. Abstract. There are two methods used by economic theory: the hypothetical-deductive method used principally by neoclassical economists, and the historical-deductive method adopted by classical and Keynesian economists. Both are legitimate, but, since economics is a substantive, not a methodological science, whose object is the economic system, the later method is more adequate. The hypothetical-deductive permits that the economist, starting from some basic assumptions – principally the maximizing agent –, deduces a consistent and mathematical theory, but at the cost of realism and relevance. Eventually the models so derived are useful tools but limited theories analyzing the economic system. In contrast, the historical-deductive method starts from the empirical observation of reality and the search for regularities and tendencies, as do the substantive sciences. As economics is a social science, such empirical reality is historical, formed of open systems, which only allow for incomplete and partially indeterminate models, but with superior explicative and predictive power. A central theoretical problem involving economics and, more generally, all social sciences is the choice of the preferential method of inquiry. Since Stuart Mill’s classical 1836 essay, and particularly after the neoclassical or marginalist school became dominant, a large part of the profession adopted a hypothetical-deductive method, although the more specific method for a substantive discipline is the empirical- deductive, and, for a substantive social science, a historical-deductive method. 1 Only with Keynes and macroeconomics revolution, the historical-deductive method, which had been used by the classical economists to understand the capitalist economic system and its growth on the long run, came out again to be fully used, now to build a general theory of the short term economic fluctuations and of the policies to achieve stability and growth. Soon, however, neoclassical economics, looking for the microfoundations of macroeconomics, began the process of returning to the hypothetical-deductive 1 I am assuming here a basic classification of sciences, distinguishing the ‘methodological sciences’, like mathematics and statistics which do not have an object but are methods, and the ‘substantive sciences’, which have. The later are subdivided into natural sciences, like physics or biology, and social sciences, like economics and sociology.