New Economist has some reading for Fed watchers:
New Economist: Productivity and the Fed: Why you shouldn't always trust real-time data: For those interested in the debate over monetary policy "smoothing" and problems with real-time data, the Federal Reserve's quandary over productivity during the 1990s is salutary. A key challenge facing the Fed at the time was the inconsistent messages it was getting about productivity. Incoming aggregate data initially suggested low productivity growth, but anecdotal firm-level evidence hinted at an acceleration. Which was right?
A new St Louis Fed Working Paper by Richard G. Anderson, and Kevin L. Kliesen examines this period in detail, including long and fascinating quotes from FOMC transcripts. The paper, Productivity Measurement and Monetary Policymaking During the 1990s, shows clearly how Greenspan refused to take the real-time data at face value:
Chairman Greenspan’s view of the nascent acceleration in productivity growth was formed largely by both his numerous contacts in the business sector and his abiding belief that the published aggregate data were not correctly measuring the effects of information technological innovations that businesses were claiming to have garnered.
With few exceptions, Greenspan views were discounted by both the Board’s staff and his colleagues on the Federal Open Market Committee (FOMC). But the Chairman was not dissuaded. He noted, for example, that available data suggested that the service sector had achieved no productivity gain in twenty years, an unlikely event. If this measurement was wrong, how many more were incorrect? As the discrepancies widened, in 1996 he requested that the Board’s economic research staff conduct a project to assess the accuracy of the Bureau of Labor Statistics’ published productivity figures.
Although that study confirmed the picture painted by the then-current data, subsequent data revisions changed the picture. Later published aggregate data converged to the more rapid growth suggested by the Chairman’s anecdotal, firm-level data.
The authors conclude:
Our analysis highlights the difficulties in formulating monetary policy using preliminary, incoming data. Policymakers should be - and are - wary about placing too much faith in initial estimates because data revisions often have significantly challenged the perceptions that policymakers previous held in “real time.”
Recommended reading for all Fed watchers.