THE USE AND ABUSE OF REAL-TIME AND ANECDOTAL INFORMATION IN MONETARY POLICYMAKING
EVAN F. KOENIG
Vice President and Senior Economist
Federal Reserve Bank of Dallas
Overview
The main message of this paper is that policymakers should not necessarily take official government statistics at face value, and should be open to alternative sources of information, including anecdotal reports and surveys. Specific instances are cited where official statistics have been misleading and where anecdotal or survey information has provided early warning of important changes in the economy.
Policymakers know not to put their faith in data series that are subject to large revisions. However, they often fail to recognize the extent to which the forecasts and policy advice they receive comes from models that are estimated and evaluated ignoring revisions, and whose performance, therefore, is at once sub-optimal and overstated. In much the same vein, policy rules that seem to perform well in after-the-fact evaluations often perform poorly in real time.
Recent research has shed light on how forecasting relationships are properly estimated when data are subject to revision, and has demonstrated that the payoff to correct estimation is often substantial. Archival requirements are not as onerous as might be expected, and historical vintage data sets have become more readily available.
The extent to which anecdotal reports and qualitative surveys are useful supplements to official statistical releases is underappreciated. Such reports and surveys are often more timely than official statistics. Moreover, respondents seem to filter out some of the short-term noise that makes economic turning points difficult to recognize in real time. Of course, skill and care are required when interpreting qualitative information just as much as when interpreting quantitative data. It helps if the qualitative information is collected from a large number and variety of sources, and if continuity of sources is maintained over an extended period. A geographically decentralized institutional structure, like that of the Federal Reserve, facilitates the flow of anecdotal information to policymakers.
Financial asset markets are forward-looking. Financial asset prices are available almost continuously and are not subject to revision. Consequently, these prices might seem to be ideal forecasting tools. Unfortunately, however, the links between asset prices and the real economy are not always straightforward, and the policy expectations implicit in asset prices are sometimes unrealistic.