How Do Data Revisions Affect the Evaluation and Conduct of Monetary Policy?

Article excerpt

Many economic data series are revised as more comprehensive information becomes available and as methodologies improve. Even the latest available data are subject to uncertainty, and at some point historical data may be replaced by more accurately measured observations. Because monetary policy decisions are made with an eye to the state of the economy, data uncertainty complicates the evaluation and conduct of monetary policy.

Revisions to data series complicate the evaluation of historical policy actions. Policy actions taken based on data available at the time may differ considerably from recommendations based on revised data (Orphanides 2001). Policy settings that seemed appropriate when they were made may be regarded as mistakes when viewed with the revised data. Thus, to understand the policy concerns of the past, it is important to know the data policymakers were observing at the time (Runkle; Croushore and Stark 2000).

A related consequence of data revisions is that policymakers today must make decisions based on data that they know are noisy or imprecise measures of activity. Since different data series are revised to different extents, knowing the justification and properties of historical revisions can help policymakers distinguish important economic signals in data fluctuations from noise. Data subject to smaller revisions might be regarded as less noisy and more reliable. Consequently, when making decisions policymakers might choose to put more weight on such series.

This article focuses on revisions to data that policymakers often examine when assessing monetary policy options. While other studies have looked at the impact of data revisions on monetary policy, this article is the first to examine the policy implications of revisions in two widely used benchmarks of resource utilization-the Congressional Budget Office (CBO) estimates of potential output and the natural rate of unemployment. The article is also the first to consider how data revisions affect policy decisions through changes in estimates of the equilibrium real rate of interest.

The article finds that revisions to data can lead to policy regretinstances when revised data may suggest alternative actions would have been preferable to those taken. Based on this finding and analysis in other studies, the article recommends making policy less sensitive to economic indicators that are subject to large revisions. The first section of the article addresses what data revisions are relevant for monetary policymakers. The second section reviews the timing and magnitude of historical revisions to a key set of policy-relevant indicators. The third section illustrates how data revisions complicate the evaluation and conduct of monetary policy. The fourth section reviews the strengths and weaknesses of proposals on how to minimize the complications of data revisions when setting monetary policy.


Most macroeconomic data series are revised. Although financial data, such as bilateral exchange rates and security prices, generally are not revised, measures of real economic activity and aggregate prices typically are. This article analyzes measurements of activity and inflation that are representative of the economic variables typically encountered in studies on the conduct of monetary policy.

The choice of series is motivated by recent literature drawing on Taylor (1993) that uses policy rules to recommend a target level for the federal funds rate based on the equilibrium real interest rate, a measure of economic activity, and the deviation of inflation from the policymakers' inflation goal.1 Such policy rules are convenient for highlighting the implications of data uncertainly but do not capture the complications of the actual policymakmg process. Nevertheless, as argued by Taylor (1993) and Svensson, the rules may provide useful benchmarks for thinking about policy in a complex world. …