Dynamic Factor Analysis for Measuring Money

Technological innovations in the financial industry pose major problems for the measurement of monetary aggregates. The authors describe work on a new measure of money that has a more satisfactory means of identifying and removing the effects of financial innovations. The new method distinguishes between the measured data (currency and deposit balances) and the underlying phenomena of interest (the intended use of money for transactions and savings). Although the classification scheme used for monetary aggregates was originally designed to provide a proxy for the phenomena of interest, it is breaking down. The authors feel it is beneficial to move to an explicit attempt to measure an index of intended use. The distinction is only a preliminary step. It provides a mechanism that allows for financial innovations to affect measured data without fundamentally altering the underlying phenomena being measured, but it does not automatically accommodate financial innovations. To achieve that step will require further work. At least intuitively, however, the focus on an explicit measurement model provides a better framework for identifying when financial innovations change the measured data. Although the work is preliminary, and there are many outstanding problems, if the approach proves successful it will result in the most fundamental reformulation in the way money is measured since the introduction of monetary aggregates half a century ago. The authors review previous methodologies and describe a dynamic factor approach that makes an explicit distinction between the measured data and the underlying phenomena. They present some preliminary estimates using simulated and real data.