Statistical Testing of Business-Cycle Theories

CAN statistical inference based on probability theory be applied to economic time series? Textbooks on business cycles often answer this question in the negative, after some superficial remarks about the difficulties of interdependence between observations, nonrandom behavior of time series, etc. In his Business Cycles Professor Schumpeter has tried to dig much deeper.' His analysis gives us for the first time, I believe -some real clues toward understanding why these problems have been and remain the constant worry of economists. The present writer has had the opportunity of discussing these problems at length with Professor Schumpeter himself. The brief analyses and comments presented below are, I might say, a direct outgrowth of those interesting discussions. But this, I should add, does not imply a conformity of opinions on all points.