Gaining an Edge via Textual Analysis of FOMC Meetings

The article “Gaining an Edge via Textual Analysis of FOMC Meetings” first appeared on Alpha Architect Blog.

Excerpt

Real-time price discovery via verbal communication: Method
and application to Fedspeak

• Roberto Gómez-Cram, Marco Grotteria
• Journal of Financial Economics
• A version of this paper can be found here
• Want to read our summaries of academic finance papers? Check out our Academic Research Insight category

What are the Research Questions?

How investors understand and use central bank communications, aka FEDSPEAK, is oftentimes cryptic and difficult to analyze.  This study attempts to provide some clarity to this issue by applying textual analysis to both high-frequency price and communication data, to focus on episodes whereby stock price movements are identifiable and on investors’ reactions to specific sentences communicated by the Fed. The sample included 41 press conferences that occurred from April 2011 to January 2020. Each conference consisted of an opening statement and a question and answer period.  The news was also extracted from the FOMC meeting statements.  The authors describe it best:

We first scrape the videos of the Fed chairman’s post-meeting press conference. Then we convert the audio into interpretable text and time-stamp it at one-second intervals. Next, we align high-frequency financial data with the exact words pronounced in each moment. This approach allows us to match a specific message with the market’s response.

What are the Academic Insights?

Using Fed funds futures, Eurodollar futures, stocks, and forex, the authors report the following results:

1. For changes in statement-related minutes, trading volume and volatility increase significantly, especially those in which the chairman discusses the future. Thus, changes or “shocks” provide valuable information to the market about the future path of monetary policy.
2. Implied volatility from options on Eurodollar futures is related to significant reductions in interest rate uncertainty for policy statement changes. From press conferences, the reductions in uncertainty extend to longer time horizons.
3. However, price changes for almost all of the securities analyzed, were comparable whether the “shock” comes from the press conference or the statement minutes. With the exception of fed fund futures, the results were statistically significant with similar slope estimates across all asset classes (from 0.2 to 0.4).  When the same analysis was repeated for a “placebo” event (FOMC without a press conference) and a similar time window, there was no evidence of positive price changes.
4. The strong, positive relationship in price changes is concentrated when the FOMC event coincides with a larger dispersion of analyst forecasts about the policy decision and/or when FOMC events coincide with a VIX above its historical average.
5. The results of an empirical test of a trading strategy based on the observed patterns are presented in Figure 3 below. For the SPY, Eurodollar futures, and the EUR/USD exchange rate, the 30-minute return around the FOMC statement release is treated as a trading signal. At the beginning of the press conference, a long position is taken if the price increased when the statement was released. Otherwise, a short position is taken.  Note that the outperformance begins 10 minutes into the Q&A session and tends to stabilize 20 minutes into the session for the SPY but steadily grows to the end of the session for the other two securities.

Why Does it Matter?

The unexpected contribution of this paper is the novel use of textual analysis to study how investors form expectations from events that are available for public consumption.  The authors align information extracted from videos of the FOMC press conference and time-stamp each word such that they can be matched to high-frequency price data for several financial securities. The authors lay out, in great detail, the methodology that adds a time dimension to textual analysis and allows a harmonization of linguistics with market prices.