Did Harvard barometers allow for the prediction of the 1929 Stock market crash

Abstract

The Harvard barometers were an attempt to analyse and predict the business cycles, which took place in the 1920s. An initiative from the Harvard Economic Service (HES), it was one of the first and more important instrument used to try to understand the sequence in the economic fluctuations. This paper reconsiders the accepted position about the Harvard barometers, that using them it was impossible to predict the 1929 Depression. I arrive at a different conclusion. Based on the data from the ABC curves in August 1929, and with an available econometric methodology at that time, it would have been possible to forecast the fall in speculation, as defined in the curve A, whereas the fall in business (B), and in monetary and credit conditions (C) were unpredictable. The stock market crash could have been anticipated. The HES stated that curve A precedes B, and then C. This is not detected. This paper makes use of the harmonic analysis by breaking down series in sinusoidal curves. Taking into account this prediction, this work analyses if aggregation was the factor producing the perceived regularities. The conclusion is negative: aggregation did not produce those cycles, they were in the original data.