How should one conclude whether the data have come in stronger, weaker, or as expected?
At each Federal Open Market Committee (FOMC) meeting the members make decisions on monetary policy. Since monetary policy is forward looking, changes in those policies are expected only if unexpected information on the state of the U.S. economy has arrived since the previous meeting. If the economic data come in stronger (weaker) than expected at the previous FOMC meeting, then the members might enact tighter (looser) monetary policy. So, clearly, predicting whether the Fed will change its policy depends on having an understanding of whether the incoming data have been stronger or weaker than expected.
Understanding the data at this depth is complicated because many macroeconomic variables are released between one FOMC meeting and the next and it is unlikely that all of the variables will turn out uniformly better or worse than expected. Some variables or economic indicators will be stronger and some will be weaker. How should one conclude whether the overall data have come in stronger, weaker, or as expected? One approach is for Fed watchers to use their own subjective judgment to characterize the data. An alternative approach is to use a quantitative measure constructed by economists at Citigroup. Their “Economic Surprise Index” is constructed daily by taking a weighted average of the data “surprises” (actual releases versus Bloomberg survey median forecasts) observed over the past three months. The index focuses on key data releases, such as the monthly jobs estimate and the monthly change in the consumer price index, as well as other data releases. Older surprises are discounted relative to more recent surprises to prevent the index from becoming stale. A value greater (less) than zero denotes stronger- (weaker)-than-expected data, whereas a value near zero indicates that the data have been coming in as expected.
The chart shows a plot of the Economic Surprise Index since 2003. As expected, the index is somewhat volatile with a mean of roughly zero. The series is modestly persistent with an autocorrelation coefficient of 0.66—hence there are fairly long swings within which the series stays positive or negative. This persistence casts doubt on the usefulness of the index because it suggests that the surveyed economists exhibit a degree of excessive “stickiness” on their views of the economy as a whole: They tend to be overly optimistic for too long and then overly pessimistic for too long.
Another reason to doubt the usefulness of the index relates to how the weights are constructed. First, the weights are based on the historical impact of each data surprise on the U.S. exchange rate rather than on the variables that most concern the FOMC, including unemployment and inflation rates. Second, the weights are not directly designed for policy-oriented horizons of one to two years ahead.
Despite these weaknesses, the index does have some predictive content for changes in monetary policy as measured by changes in the federal funds rate across FOMC meetings. The correlation between the value of the index observed on the day of the meeting and the change in the effective funds rate is 0.18 and 0.25 at the current FOMC meeting and the following meeting, respectively. Moreover, the chart indicates that the major quantitative easing announcements did in fact occur at times when the index had fallen sharply. In short, while the index likely is not of value to the FOMC as a means of tracking unexpected strengths or weaknesses of the economy, Fed watchers might find it modestly useful for tracking FOMC policy decisions.