"...In 1991, the Scandinavian Journal of Economics organized a conference ëNew Ap- proaches to Empirical Economicsíin which Lawrence Summers contributed the pa- per ëThe ScientiÖc Illusion in Empirical Economicsí(Summers, 1991). He argued in the paper that empirical economics has exerted little ináuence on the development of economic theory and provided little new insight on economic mechanisms. As an illustration, he mentioned empirical analysis of representative agent models, which involves estimating a few deep parameters characterizing preferences and technology. He contrasted this analysis with sophisticated statistical techniques that impose min- imal theoretical constraints on the data, as exempliÖed by a Vector AutoRegressive (VAR) model ‡ la Sims. Summers argued that both approaches have been unable to explain the inherently richer and more complicated macroeconomic reality. He concluded that less formal examination of so-called stylized facts (correlations, mean growth rates, graphs) has resulted in more fruitful economic research.
Almost twenty years have passed since Summerís critique and the INET confer- ence seems like a good opportunity to revisit his main arguments. Has the marriage between theory and evidence improved? Has empirical economics become more enlightening?.."
This paper, which is prepared for the Inagural Conference of the Institute for New Economic Thinking in Kingís College, Cambridge, 8-11 April 2010, questions the preeminence of theory over empirics in economics and argues that empirical econometrics needs to be given a more important and inde- pendent role in economic analysis, not only to have some confidence in the soundness of our empirical Read more