Abstract
The purpose of this paper is to empirically test the relationship between velocity growth and anticipated and unanticipated money growth and its variability. In particular, by decomposing actual monetary growth into anticipated and unanticipated components, the study investigates whether knowledge of past variability of anticipated and unanticipated monetary growth improves the prediction of future velocity growth beyond predictions that are based on past velocity growth alone. This is the empirical definition of Granger causality. Comparasions are made, in all tests, among simple sum and Divisia aggregates (of M1, M2, M3, and L) and with Spindt's monetary velocity (MQ) aggregate.