Picking Up Pennies in Bond Markets on FOMC Days

Summary

• A new Journal of Financial Economics paper explains how Blue-Chip professionals’ GDP forecast revisions can give early insights into bond market movements on FOMC days.
• The authors find upward revisions in year-ahead GDP forecasts predict a rise in bond yields in a 30-minute window around the FOMC announcements, especially during crises.
• A strategy that goes short rates during the 30-minute window on months when GDP forecasts are revised upwards and long rates after downward revisions yields a mean annualised return of 15bps – not bad for four hours’ trading a year!

Introduction

Investors have kept their eyes glued more closely to FOMC meetings since the GFC. In relative terms, bond yields on US 1y now regularly move as much as 10% in value on Fed meeting days (Chart 1). And, in absolute terms, US equities move on average 1%, up from 0.7% pre-GFC (Chart 2). This increased to 1.9% in 2022.

With increasingly volatile asset prices on FOMC meeting days, early insight into their direction would be valuable. And such early signals exist, according to the authors of a new Journal of Financial Economics paper. They show forecast revisions of Blue-Chip professionals (and to a lesser extent yield spreads and uncertainty measures) have predictive power over bond movements in 30-minute windows around FOMC meetings.  

This statistical pattern implies investors consistently misprice bond markets ahead of Fed announcements. Specifically, when year-ahead GDP growth forecasts are revised upwards pre-FOMC meeting, the market underestimates how contractionary the subsequent FOMC announcement will be. So bond yields move predictably higher. This phenomenon leads to a profitable trading strategy, especially during crises.

The Paper in a Nutshell

The change in interest rate futures and bond yields around FOMC announcements has become a popular measure of monetary policy news. It assumes before the FOMC announcements, the prices of these assets fully incorporate all public information about what the Fed will say. So, changes in prices after the event capture ‘surprises’ announced by the Fed.

In essence, the paper challenges that view. If movements in bond yields are predictable, investors must be missing valuable information pre-FOMC. The authors argue Blue-Chip professionals have that information. And, by altering their GDP forecasts, these professionals provide a good signal on the tone of the subsequent FOMC announcement, helping predict bond movements.

Data

The authors begin by identifying their monetary policy news/shock variable. They do it in two ways, but I focus on one for simplicity because the results are the same. They take the average change in five interest rate futures – 1m, 3m FF futures and 3m Eurodollar interest rates at two-, three- and four-quarter horizons – from 10 min before the FOMC announcement to 20 min after it.

Next, they collect survey data from Blue Chip Economic Indicators. These surveys run from the fifth to the seventh day of every month, asking economists at large American firms about their forecasts of key economic variables. These forecasts are then released to subscribers on the 10^th day of every month.

The authors use survey data on forecasts of quarter-to-quarter GDP growth for the current quarter and the next three quarters. They then calculate the monthly revisions in these numbers.

For example, if the current quarter forecast is March, they take the difference between March and February current quarter forecasts. However, for the first month in a quarter, they take the difference between the current quarter forecast (e.g., in April) and the quarter-ahead forecast in the previous month (e.g., in March).

Methodology

The authors try to show the extent to which revisions in economic forecasts can explain changes in interest rates during the 30-minute window around FOMC announcements. For this, they run a simple regression, where the monthly revisions are on the year-ahead GDP forecast.

For example, for an FOMC announcement on 13 August 2002, they use the Blue-Chip forecast revision available as of 10 August 2002. The year-ahead forecast period is July 2002–June 2003. The forecast revision captures a change in GDP growth expectations from 5-7 July to 5-7 August.

As an extension, they then remove these revisions and replace them with changes in key financial and macroeconomic variables, such as 6m Treasury spreads and the unemployment rate.

Forecast Revisions and Interest Rates

The results confirm that forecast revisions from the Blue-Chip survey predict changes in interest rates during FOMC announcements. An upward revision in the year-ahead average quarterly growth forecast by 1pp predicts a 7bp rise in interest rates – which is a contractionary monetary policy shock (Chart 3).

Most of the predictive power is from revisions in the next-quarter-ahead growth forecast, with revisions to the last quarter least important. Also, the predictive power is highest during crises (November 2001, October 2008 and December 2008) and during the zero-lower bound period from January 2009 to December 2015.

Forecast Revisions and Other Yields

The predictive power may not just be over the five interest rate futures used above. So, the authors see whether GDP forecast revisions can also predict changes in other yields. They can, especially those with 6- to 12-month maturities, such as 6m Fed Funds futures and 2-quarter Eurodollar futures. They do, however, also predict changes in 2y and 10y yields in the 30min window around FOMC announcements.

What Drives the Results?

Building on another paper’s analysis, the authors decompose the interest rate changes into four components: growth news, monetary news, hedging premium news and common premium news. Only the monetary news component is significant, implying investors typically underreact to monetary policy news before FOMC announcements.

Why? Investors overweight the public signals that the Fed releases during FOMC meetings. They are less likely to give equal attention to economic conditions as professional forecasters are, and therefore do not fully price in what the Fed will do – they may get 90% of the way, but tactically leave 10% to be confirmed during the FOMC announcements.

Trading Strategy

To illustrate the economic significance of their empirical results, the authors build a trading strategy that exploits the predictability of yield changes. They use an equally weighted portfolio of 1m, 3m, and 6m fed funds futures and 2Q, 3Q, and 4Q Eurodollar futures. This portfolio is the short-term level yield factor.

Their strategy bets on an interest rate increase at an FOMC announcement if Blue-Chip professionals have revised their one-year GDP expectations upwards in the most recent release. If Blue-Chip professionals recently revised their GDP expectations downwards, the strategy bets on an interest rate decrease.

The strategy yields a cumulative 350bps gain over 20 years, equivalent to a 15bps mean annualised return with a SR of 0.75 (Chart 4). Just like picking up pennies, but without the large steamroller!

Bottom Line

In an environment dominated by macroeconomic uncertainty, many look towards the Fed for clarity on the state of the economy and the direction of monetary policy. This has amplified market movements on FOMC announcement days, presenting an opportunity for investors in the know.

This paper offers a sensible early warning signal pre-FOMC meetings: if professional forecasters are revising their expectations ahead of Fed announcements, note their direction. On average, investors do not fully price these changes in, so there could be a chance of making money!