Economic news typically leads to adjustments in the prices of financial assets, particularly bonds and stocks. Basic economic thinking suggests that bond yields will rise if announcements are stronger than expected, and that exchange rates will move in response to inflationary pressures. In fact, we find that the sign and magnitude of such responses are generally consistent over different intraday intervals, with stock prices responding more flexibly. We also identify economically and measurably persistent responses for several indicators, with bond yields showing the strongest responses and stock prices the weakest.
The standard approach to measuring the impact of news relies on survey data collected in advance of the data release, with lags of a few days or even a week or more. Such lags suggest that some information may be accumulated between the time of the survey and the actual data release, which reduces the impact per unit of true news on asset prices. In addition, the standard approach tends to overestimate the impact of news because it combines true and false news in one measure (adjusted for survey-based forecast errors).
We develop an alternative method that eliminates the problem of overlapping true and false news by regressing estimated asset price responses on both the indicator as reported and the indicator as expected at the time of the survey. Using this methodology, we find that the estimated asset price responses are significantly larger than those estimated under the standard OLS approach. In addition, the Rigobon-Sack estimates are also measurably more consistent across different intraday intervals, suggesting that their results are less sensitive to survey-based error.
About the Author
