From data to analysis

By the mid-1950s Friedman and Schwartz had assembled most of the basic U.S. data for their study. The core set of data they began with were monthly data on currency in the hands of the public from 1917 through 1944 (Schwartz and Oliver, 1947). In May 1954 Friedman reported in the National Bureau's annual report that they had extended this series back to 1907, the year of a severe banking panic that led directly to the Aldrich–Vreeland Act and appointment of the National Monetary Commission and ultimately to the Federal Reserve Act. They had also revised the series to account more accurately for the allocation of currency between public circulation and vault cash. They had produced a monthly series on commercial bank deposits from 1919 forward and were in the process of extending the series back to 1907. Comparable data for mutual savings bank and Postal Savings deposits were in the works.

A significant portion of their data compilation effort went toward the problem of estimating vault cash and deposits in banks that were not national banks (for the pre–Federal Reserve period before 1914) or members of the Federal Reserve system (from 1914 on), and converting irregular call-date data into monthly estimates. This involved interpolating to produce the “missing” monthly observations. One byproduct of this effort was that Friedman discovered a flaw in what were at the time standard procedures for using one series to interpolate between observations in another.