How could everyone have been so wrong? Forecasting the Great Depression with the railroads

Contemporary observers viewed the recession that began in the summer of 1929 as nothing extraordinary. Recent analyses have shown that the subsequent large deflation was econometrically forecastable, implying that a driving force in the depression was the high expected real interest rates faced by business. Using a neglected data set of forecasts by railroad shippers, we find that business was surprised by the magnitude of the great depression. We show that an ARIMA or Holt- Winters model of railroad shipments would have produced much smaller forecast errors than those indicated by the surveys. The depth and duration of the depression was beyond the experience of business, which appears to have believed that recovery would happen quickly as in previous recessions. This failure to anticipate the collapse of the economy suggests roles for both high real rates of interest and a debt deflation in the propagation of the depression.