The change to the playing field and character of the discipline of economics had one quite literal aspect to it. After the Second World War, the centre of gravity in economics moved to the United States, where it has remained ever since. This shift had begun in the 1930s with the emigration of academics and intellectuals, a large proportion of them Jewish, especially from Central Europe. Many of these exiles contributed to the rise of mathematical economics and econometrics in the United States. In the American context, the debate about the merits of econometrics took a different turn from that seen in Europe. The argument was specifically about the merits of empirical statistical research as against the theory-based modelling that came to fruition with the work of the Cowles Commission for Econometric Research. It was not about the merits of empirical statistical research as such. Even before the Second World War, one major strand of research in the United States was focused on statistics, an example being the Economic Service at Harvard University, which, as we have seen, constructed business-cycle barometers. The years between the wars saw the birth of a further two statistically oriented research institutes: the NBER allied to Columbia University in New York and the Cowles Commission for Econometric Research. The Cowles Commission was to grow into one of the leading institutes for econometric research in the United States. Founded in 1932 by businessman and economist Alfred Cowles, it set up shop in Chicago in the university’s department of economic and other social sciences, before moving to Yale University in New Haven in 1955. The Cowles Commission had a tense relationship with both the NBER and the economics faculty at Chicago. It is time to take a close look at the famous ‘measurement without theory’ controversy, which erupted shortly after the Second World War. Tjalling Koopmans, an economist and physicist of Dutch origin, and successor to Jacob Marschak as director of the Cowles Commission, expressed blunt criticism of an important study of business cycles by the director of the NBER, Wesley Clair

Mitchell, and his colleague Arthur Burns. Shortly after that, the econometric models developed by Lawrence Klein at the Cowles Commission came under intense scrutiny. The source of this latter expostulation was Milton Friedman, who, as a former student of Burns and a member of staff at the economics faculty in Chicago, had close links with the NBER. Even before the war, Friedman had expressed restrained criticism of the study into business cycles carried out by Tinbergen for the League of Nations, demonstrating his profound interest in empirical statistical research. In essence, Friedman raised the question of whether Tinbergen’s econometric model had actually been tested against statistical data, or whether it had, in fact, merely been adjusted as far as possible to fit the statistical data available. Unlike Keynes, Friedman worded his criticism in such a way that it suited the new playing field of the science of economics, where the boundaries are determined by mathematics and statistics. His assessment also featured a specific idea about the relationship between economic science and economic politics, a notion he expressed explicitly in his famous essay ‘The Methodology of Positive Economics’ of 1953. A discussion of Friedman’s criticism of Tinbergen and the Cowles Commission therefore serves as a prelude to a review of that essay which, along with John Stuart Mill’s essay of 1836, is among the most discussed, quoted and criticized articles on the methodology of economics. At first sight, Friedman seems to defend the view that the structure of economics – which renders up the test criteria – is no different from that of other sciences. But his defence rests on an important assumption about the underlying image of science. According to modern philosophy of science, experimental science epitomizes the image of the sciences in general, but Friedman argued that economics was not compatible with that image. He also claimed that the use of ‘the experimental method’ as a reference point in identifying good science actually created a false impression of the character of the natural sciences. This gave rise to Friedman’s famous (or notorious) ‘as if’ methodology: economic models do not represent reality, rather they are useful instruments based on largely unrealistic assumptions. This sixteen-word summary of Friedman’s position relies upon notions that have been furiously contested and subjected to minute philosophical scrutiny. What is a model? What is an economic instrument? What does it mean to say that an assumption is ‘unrealistic’? We shall see that Friedman’s methodology contains a strongly ideological message about the relationship between economic theory, empiricism and economic policy.