On the use of dispersion measures from NAPM surveys in business cycle forecasting

Abstract

Abstract Qualitative survey data on changes in production, inventory, new order, and employment collected every month by the National Association of Purchasing Managers are analysed over 1948–90. The Probability method we use generates time‐series estimates of cross‐section variabilities across firms. It is shown that these diffusion measures have additional explanatory power in the prediction of business cycle turning points.