Stock Returns Over The FOMC Cycle
University of California Berkeley and NBER
University of California Berkeley, NBER and CEPR*
First draft: April 23, 2014
Current draft: October 9, 2015
We document that since 1994 the equity premium in the US and in the rest of the world is earned entirely in weeks 0, 2, 4 and 6 in FOMC cycle time, i.e. in time since the last Federal Open Market Committee meeting. This likely reflects a risk premium for news (about monetary policy or the macro economy) coming from the Federal Reserve: (1) The FOMC calendar is quite irregular and changes across sub-periods over which our finding is robust. (2) Even weeks in FOMC cycle time do not line up with important macro releases. (3) Volatility in the federal funds market peaks during even weeks in FOMC cycle time. (4) Information processing/decision making within the Fed tends to happen bi-weekly in FOMC cycle time: The bi-weekly cycle is driven mainly by even week observations that follow board meetings of the Board of Governors. Furthermore, before 1994, intermeeting target changes were common and disproportionately took place during even weeks in FOMC cycle time. High return weeks do not line up with public information releases from the Federal Reserve or with the frequency of speeches by Fed officials. Systematic informal communication of Federal Reserve officials with the media and the financial sector is a more plausible information transmission mechanism. We discuss the social costs and benefits of this method of communication.
Stock Returns Over The FOMC Cycle – Introduction
This paper documents a striking new fact about U.S. stock returns. Over the last 20 years, the average excess return on stocks over Treasury bills follows a bi-weekly pattern over the Federal Open Market Committee (FOMC) meeting cycle. The equity premium over this 20-year period was earned entirely in weeks 0, 2, 4 and 6 in FOMC cycle time, where the FOMC cycle starts on the day before a scheduled FOMC announcement day and resets at each of the eight times the FOMC meets per year. We document that this stock return pattern reflects a large risk premium for news coming from the Federal Reserve.
The fact that stocks do well in week 0 in FOMC cycle time has recently been documented by Lucca and Moench (2015), who show that since 1994 stock returns have averaged about ½ percent over the 2pm to 2pm period prior to scheduled FOMC announcements.1 The bi-weekly pattern in stock returns over the FOMC cycle that we document here is a novel finding and appears surprisingly robust. Average excess returns are statistically significantly higher in even weeks than in odd weeks in FOMC cycle time. Furthermore, the pattern is robust across three sub-periods of this 20-year sample and is also present in stock markets outside the United States. Implied volatilities peak in a pattern consistent with the fact, showing that at least some market participants worry more about news coming out in even weeks.
Having established this fact, we present evidence that the pattern likely reflects a risk premium for news (about monetary policy or about the macro economy) coming from the Federal Reserve. First, we document that the timing of the FOMC meetings across days of the year is quite irregular; it changes across the three sub-periods over which our finding is robust, and it does not line up with reserve maintenance periods (two-week periods over which reserve requirements on banks are applied). Second, using Bloomberg macroeconomic data releases, we document that our “even week” effect in FOMC cycle time does not line up with other important macro news. Third, we show that volatility in the fed funds market peaks in even weeks in FOMC cycle time, suggesting that news about monetary policy is coming out during these weeks (this claim relies on the assumption that banks change their trading in the federal funds market in response to news about future short rates). Volatilities of both stocks and 10-year Treasuries do not show similar bi-weekly patterns indicating that what drives our main fact is not that more news comes out during even weeks in FOMC cycle time but that the type of news that comes out changes. Fourth, we provide evidence that information aggregation and decision making within the Federal Reserve tends to happen bi-weekly in FOMC cycle time. Prior to 1994, the timing of Fed decision-making can be inferred from changes made to the federal funds target rate, which was frequently changed in between scheduled meetings. Over the 1982:9-1993 period, the frequency of target changes shows distinct peaks in even weeks in FOMC cycle time. In the 1994-2013 period, inter-meeting target changes are rare. However, it is still the case that information aggregation and important policy discussions within the Fed occur in between meetings. We document a special role for board meetings of the Board of Governors of the Federal Reserve. We find that the biweekly stock return cycle is driven mainly by even week observations that follow those meetings.
See full PDF below.