View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

3
2017

The event-study activity puzzle
Jeffrey R. Campbell, Jonas D. M. Fisher, Alejandro Justiniano,
and Leonardo Melosi

Introduction and summary

1

If news of higher future interest rates reflects a shift in the stance of monetary policy, then we expect it to
lower projections for real gross domestic product (GDP) growth and raise projections for unemployment.
However, the data point to quite different conclusions. When market participants raise their estimates of
future interest rates after a meeting of the Federal Reserve’s monetary policy committee, the Federal Open
Market Committee (FOMC), private forecasters tend to raise their projections of real GDP growth and
lower their projections for unemployment. Campbell, Evans, Fisher, and Justiniano (2012) documented
this fact, and we labeled it the “event-study activity puzzle” in our NBER Macroeconomics Annual chapter
(Campbell, Fisher, Justiniano, and Melosi, 2017).
In this article, we summarize an empirical exploration of the effects of FOMC statements on financial
market expectations of future interest rates and private forecasters’ expectations of future macroeconomic
performance that appeared in Campbell et al. (2017). Our work resolves the event-study activity puzzle by
demonstrating that the FOMC moves market expectations of future interest rates in part by transmitting
its views of future macroeconomic fundamentals. When these are strong, private forecasters revise their
projections accordingly, while market participants mark up their expectations of future interest rates.

How does the FOMC transmit macroeconomic expectations?
The FOMC releases a statement after each of its meetings that announces any desired changes to the federal
funds rate and other policy tools, describes its assessment of current macroeconomic circumstances, and
possibly describes how further developments could influence its future policy choices. This discussion of
possible future actions is called forward guidance. Perhaps the most prominent example of forward guidance
from before the Great Recession was the use of “the Committee believes that policy accommodation can

Economic Perspectives

3 / 2017

Federal Reserve Bank of Chicago

be removed at a pace that is likely to be measured” in the FOMC’s post-meeting statements from May 2004
through November 2005.
After the FOMC set its target for the federal funds rate to its effective lower bound in December 2008, it
turned to forward guidance as one way to deliver additional monetary accommodation.1 As demonstrated
in the theoretical work of Krugman (1998) and Eggertsson and Woodford (2003), expansionary forward
guidance can improve current macroeconomic outcomes if it lowers households’ desired savings by increasing
their expectations of future inflation and output. According to the theory, such changes in expectations can
be induced by the FOMC making a credible commitment to keep the future federal funds rate lower than
it would otherwise do given the prevailing macroeconomic conditions. Campbell et al. (2012) called this
Odyssean forward guidance.2 After its August 2011 meeting, the FOMC stated that macroeconomic conditions
“are likely to warrant exceptionally low levels for the federal funds rate at least through mid-2013.”3 Similarly,
it stated after its December 2012 meeting that interest rates would remain exceptionally low “at least as
long as the unemployment rate remains above 6-1/2 percent, inflation between one and two years ahead is
projected to be no more than a half percentage point above the Committee’s 2 percent longer-run goal,
and longer-term inflation expectations continue to be well anchored.”4 Both statements could be viewed
as Odyssean forward guidance.

2

While the theoretical possibility for appropriate forward guidance to improve current macroeconomic
outcomes is clear, the evidence does not unambiguously support this conjecture. Ben Zeev, Gunn, and
Khan (2015) and D’Amico and King (2016) both use econometric models and money market futures prices
to identify unexpected movements in expected future interest rates caused by FOMC forward guidance;
they both find that output and inflation rise presently and in the future when expected future rates fall.
This is exactly as the standard New Keynesian model predicts when changes to expected future interest rates
represent exogenous adjustments to the stance of monetary policy. On the other hand, Campbell et al. (2012)
measure unexpected movements in future interest rates with changes in money market futures rates on
the dates of FOMC meetings. When they relate revisions in private sector forecasts of output growth and
unemployment to these “forward-guidance shocks,” they find that decreases in future expected interest
rates lower expectations of output and raise expected unemployment. This is the event-study activity puzzle
mentioned above.

Quantifying the puzzle
To quantify the event-study activity puzzle, we related changes in private sector forecasts for GDP growth
and unemployment over months containing an FOMC meeting with the change in the expectation of the
federal funds rate four quarters into the future, on the day of the FOMC announcement. By construction,
this event-study of how monetary policy announcements affect expectations of future monetary policy
controls for all information available to market participants at the close of business on the previous day.
Since FOMC announcements are usually the most important financial news on the day they occur, they
should dominate our measured changes in expectations of interest rates. Our measures of private sector
forecasts come from the Blue Chip Economic Indicators, a monthly survey of private forecasters. We use
its “consensus” forecast, which averages all submitted forecasts. Its respondents provide quarterly forecasts for real GDP growth and the unemployment rate for the current and next calendar years. Therefore,
the survey reports private forecasts for the current and next four quarters within one month of every
FOMC meeting. For each of these observables, we estimate the coefficients of a simple linear model,
∆ m et = α + β∆ d it4 + ε t

Economic Perspectives

3 / 2017

Federal Reserve Bank of Chicago

Here, ∆met is the change in the Blue Chip forecast for some variable over the month containing a particular
FOMC meeting, ∆ d it4 is the change in expectation of the federal funds rate four quarters hence on the day
of the FOMC meeting, and εt is a residual that accounts for all movements in ∆met that are unrelated to
∆ d it4 . We estimated the model’s unknown coefficients α and β using data from all FOMC meetings between
1994 and 2010 inclusive.5
For each of the private expectations, we estimated the unknown coefficients α and β using the method of
ordinary least squares. This method makes the sum of the squared implied values for εt as small as
4
possible. Table 1 reports the estimated coefficients multiplying ∆ d it . When these are positive, the data
indicate that increases in the interest rate expected four quarters in the future are associated with increases
in the given expectation. Of course, sampling error could make any of our coefficient estimates non-zero
even though the coefficient’s true value is zero. For each regression, we tested the null hypothesis that the
true value of β equals zero against the alternative that it is non-zero using methods described in Campbell
et al. (2017). When the test rejects the null hypothesis at the conventional levels of statistical significance
of 5 percent and 1 percent, we mark the associated estimate of β with ** and ***, respectively. A smaller
level of statistical significance means that the chances of a model with a true coefficient of zero generating the estimated non-zero coefficient are lower. Next to each estimated coefficient is the model’s R2
measure. This varies between 0 and 1 and gives the fraction of variance in ∆met that the model assigns to
movements in ∆ d it4 .

3

The results in table 1 illustrate the event-study activity puzzle. Raising the expectation of the federal funds
rate four quarters hence raises private sector forecasts of GDP growth and lowers unemployment forecasts for
the next four quarters. For example, a 1 percentage point increase in ∆ d it4 raises the expectation of annualized
GDP growth in the next quarter by 66 basis points and lowers the expectation of next quarter’s unemployment rate by 21 basis points. Only the changes in expectations for the next quarter are statistically significant at conventional levels. For expected GDP growth, the coefficients’ estimated magnitudes at horizons
of two to four quarters vary from 27 to 6 basis points. For unemployment, the coefficient estimates for
those longer horizons are comparable to those for the next quarter, but the coefficients are measured relatively
imprecisely. The R2 measures in table 1 all indicate that most variance in expectations revisions is due to
factors other than changes in expected interest rates. That is, these estimates indicate that FOMC forward
guidance was not a key driver of private expectations during the sample period.
TABLE 1

			

Change in private forecast of
GDP growth
Next quarter
Two quarters hence
Three quarters hence
Four quarters hence
Unemployment
Next quarter
Two quarters hence
Three quarters hence
Four quarters hence

The event-study activity puzzle
β (coefficient multiplying ∆ d it4 )
0.66***
0.27
0.07
0.06
–0.21**
–0.19
–0.17
–0.18

R2
0.03
0.01
0.00
0.00
0.05
0.02
0.02
0.02

Note: ** and *** indicate that a test rejects the null hypothesis at the 5 percent and 1 percent level, respectively.
Sources: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System and Blue Chip
Economic Indicators.

Economic Perspectives

3 / 2017

Federal Reserve Bank of Chicago

Campbell et al. (2012) speculated that their results arose from a failure of their measured shocks to represent
truly exogenous changes in the stance of monetary policy. In general, the FOMC could surprise market
participants in two distinct ways. First, it might truly change the stance of monetary policy based on its
members’ judgments about commonly observed macroeconomic fundamentals. Second, it could signal that
it has received information about future macroeconomic developments that is unavailable to market participants.
Although FOMC participants have very little access to macroeconomic data that are unavailable to the
public, the Federal Reserve System’s staff might give the FOMC proprietary interpretative information. If
this is so, then transmitting information that the economy would “run hot” in the near future, for instance,
would lead market participants to increase their forecasts of both interest rates and output. Campbell et al.
(2012) labeled such forward guidance that simply reports information on future macroeconomic developments
and discussed possible FOMC reactions to those developments as Delphic.6 If most FOMC forward guidance is
Delphic, then the event-study activity puzzle can be understood as the optimal reactions of market participants
and private forecasters to prognostications about future macroeconomic developments in FOMC statements.
Although this Delphic resolution to the event-study activity puzzle seems plausible, Campbell et al. (2012)
provided no evidence to support it. In Campbell et al. (2017), we subjected this hypothesis to empirical
scrutiny using measures of the FOMC’s private information about short-run macroeconomic performance.
For this, we proceeded in two steps. First, we inferred the change in interest rates that can be directly explained
by these measures of private information, and we attributed this estimate to FOMC short-run Delphic forward
guidance. Second, we quantified the extent to which short-run Delphic forward guidance explains the eventstudy activity puzzle by estimating how this measure influenced changes of private sector expectations of
GDP growth and unemployment.

4

We constructed our measures of FOMC private information by subtracting consensus private sector
forecasts for Consumer Price Index (CPI) inflation, real GDP growth, and the unemployment rate for the
current and next four quarters from analogous forecasts prepared by Federal Reserve Board staff for the
FOMC. These forecasts are made available to FOMC participants (Governors of the Federal Reserve
System and the 12 Reserve Bank presidents), but they are kept out of the public domain for five years
after their creation. Therefore, any measurable effect these have on innovations to financial markets’
expectations of future interest rates should operate only through the content of the FOMC statement.
To examine the influence of these forecast differences on changes in financial market expectations of the
federal funds rate, we estimated the coefficients of a simple linear model:
∆ d it4 = µ + γ y dty + γ u dtu + γ π dtπ + η t .
Here dty , dtu , and dtπ represent the differences between the Board staff and Blue Chip forecasts for output
growth, unemployment, and inflation, respectively. These include differences for the current meeting and
from the meeting that preceded it. We included this seemingly stale information because FOMC participants
might require time to form the consensus about the meaning of new information required for its discussion
in a meeting statement.7 We estimated the model’s unknown coefficients γy, γu, and γπ with the method of
ordinary least squares, as described earlier.
Table 2 gives the results of this estimation. For the sake of parsimony, it reports the results of exclusion
tests rather than the estimated coefficients themselves. These test the null hypothesis that the coefficients
multiplying each variable in a group all equal zero. For example, the exclusion test for all FOMC private
information about GDP growth uses the null hypothesis that all elements of γy equal zero. Just as before,
the superscripts ** and *** indicate that a test rejects the null hypothesis at the 5 percent and 1 percent level,
respectively.8 The exclusion test rejects the null hypothesis that all of the forecast differences are irrelevant at

Economic Perspectives

3 / 2017

Federal Reserve Bank of Chicago

TABLE 2

Revelation of Greenbook forecasts:
Exclusion tests
			
Variables excluded
All
Current meeting’s
Previous meeting’s
GDP forecast differences
Unemployment forecast differences
Inflation forecast differences

Test statistic
33.33**
11.08
19.34***
19.25***
7.79
1.78

Note: ** and *** indicate that a test rejects the null hypothesis
at the 5 percent and 1 percent level, respectively.
Sources: Authors’ calculations based on data from the Board
of Governors of the Federal Reserve System and Blue Chip
Economic Indicators.

the 5 percent level. Interestingly, the test does not
reject the null hypothesis that the forecast differences
associated with the current meeting are irrelevant.
Instead, it is the previous meeting’s forecast
differences that help explain the change in expected
future interest rates following the current meeting.
As suggested earlier, we interpret this result as
reflecting the time required for FOMC participants
to process new information. Among the three macroeconomic variables included, the exclusion tests
indicate that it is differences between the Board staff
and Blue Chip forecasts for near-term GDP growth
that have the most influence on financial market
participants’ expectations of future interest rates.9

These empirical results strongly suggest that the
FOMC routinely transmits the Board staff’s view
of future macroeconomic conditions to the public through FOMC statements. Additionally, they provide
the ability to control for these transmissions when assessing the influence of statement-induced changes in
expected future interest rates on private macroeconomic forecasts. For this, we use the estimated values of
γy, γu, and γπ to construct the change in private interest rate expectations expected given the differences
between the Board staff and Blue Chip forecasts. Call this estimate of the FOMC’s short-run Delphic
forward guidance fˆt :
π π
fˆt ≡ µˆ + γˆ y dty + γˆ u dtu + γˆ dt .

Here, the hats indicate that the linear model’s coefficients are estimated from data. With this, we can write

5

∆ d it4 = fˆt + ηˆ t .
The model’s estimated residual term is ηˆ t . By construction, its correlation with fˆt in our sample equals
zero. It represents all influences on ∆ d it4 that are not due to short-run Delphic forward guidance. These
include long-run Delphic forward guidance and any Odyssean forward guidance.
We can now assess how these two components of ∆ d it4 shape revisions to private sector expectations.
There is no good ex ante reason to expect the short-run Delphic forward guidance in fˆt to influence
private macroeconomic expectations in the same way as the various messages embodied in ηˆ t do. To
account for these possible differences, we can generalize the linear model used to create table 1 by
allowing these two components of ∆ d it4 to have distinct coefficients. That is,
∆ m et = α + β f fˆt + β η ηˆ t + ε t .
Table 3 reports the results of estimating this model with the method of least squares. Each row reports the
model’s estimated coefficients and R2 measure for one macroeconomic expectation. The superscripts **
and *** indicate the statistical significance of the coefficient estimates at the 5 percent and 1 percent level,
respectively. The estimates clearly indicate that the FOMC’s short-run Delphic forward guidance moves
interest rate expectations and projections of GDP growth in the same direction. A 1 percentage point
increase in fˆt is associated with a 2.88 percentage point increase in expected GDP growth next quarter, a
1.75 percentage point increase two quarters hence, and a 0.66 percentage point increase three quarters

Economic Perspectives

3 / 2017

Federal Reserve Bank of Chicago

TABLE 3

Blue Chip forecasts’ responses to decomposed monetary policy
			
Delphic component (fˆt)

Residual (η̂t)

R2

GDP growth
Next quarter
Two quarters hence
Three quarters hence
Four quarters hence

2.88**
1.75***
0.66**
0.29

–0.01
–0.18
–0.11
–0.01

0.12
0.14
0.07
0.02

Unemployment
Next quarter
Two quarters hence
Three quarters hence
Four quarters hence

–0.88***
–1.08***
–1.28***
–1.18***

–0.01
0.09
0.16
0.12

0.21
0.19
0.20
0.17

Note: ** and *** indicate that a test rejects the null hypothesis at the 5 percent and 1 percent level, respectively.
Sources: Authors’ calculations based on data from the Board of Governors of the Federal Reserve System and Blue Chip
Economic Indicators.

hence. All of these responses are statistically significant at conventional levels. Similarly, the FOMC’s
short-run Delphic forward guidance moves expected future interest rates and unemployment projections
in the opposite directions. For example, a 1 percentage point increase in fˆt decreases the expected unemployment rate in four quarters by 1.18 percentage points. The R2 measures for these linear models are
between 0.17 and 0.21. That is, the FOMC’s short-run Delphic forward guidance accounts for about
one-fifth of the monthly revisions to unemployment expectations.

6

The estimated residual term ηˆ t should be free of (measurable) short-run Delphic forward guidance. The estimated coefficients indicate that this is indeed so. All but one of them has the “intuitive” sign associated with
exogenous changes to the stance of monetary policy. That is, these estimates are consistent with the prediction
that expectations of economic activity fall when the FOMC signals a tightening in future monetary policy
beyond what the private sector would ordinarily forecast given the incoming macroeconomic data. However,
they are relatively small in magnitude and none of them are statistically significant. Therefore, there remains
considerable scope for future research to investigate this hypothesis.

Conclusion
In Campbell et al. (2017), we present a more detailed analysis of these results, including a discussion of
whether forward guidance substantially affects Treasury term premiums. Overall, we find that accounting
for the FOMC’s transmission of short-run Delphic forward guidance in its meeting statements resolves the
event-study activity puzzle of Campbell et al. (2012). This result is important in itself for designing central
bank communications policy, and it clears the way for future research into Odyssean forward guidance.

Economic Perspectives

3 / 2017

Federal Reserve Bank of Chicago

NOTES
In a speech at the London School of Economics on January 13, 2009, Chairman Ben Bernanke was explicit
about the value of forward guidance when policy is constrained by the effective lower bound: “Although
the federal funds rate is now close to zero, the Federal Reserve retains a number of policy tools that can
be deployed against the crisis. One important tool is policy communication. Even if the overnight rate is
close to zero, the Committee should be able to influence longer-term interest rates by informing the public’s
expectations about the future course of monetary policy. To illustrate, in its statement after its December
meeting, the Committee expressed the view that economic conditions are likely to warrant an unusually low
federal funds rate for some time. To the extent that such statements cause the public to lengthen the horizon
over which they expect short-term rates to be held at very low levels, they will exert downward pressure
on longer-term rates, stimulating aggregate demand.” Read the complete text of the speech at
https://www.federalreserve.gov/newsevents/speech/bernanke20090113a.htm.
1

Odyssean forward guidance consists of central bankers’ statements that bind them to future courses of action.
Just as Odysseus bound himself to his ship’s mast so he could enjoy the sirens’ song without succumbing
to the temptation to drown himself while swimming toward them, a central banker can improve economic
outcomes by publicly committing to a plan that uses expectations of suboptimal future outcomes to improve
current economic conditions.
2

See https://www.federalreserve.gov/monetarypolicy/fomcminutes20110809.htm.

3

See https://www.federalreserve.gov/monetarypolicy/fomcminutes20121212.htm.

4

Our sample begins with 1994 because the FOMC issued statements only intermittently before then, and it
ends in 2010 because the Greenbook forecasts we used to measure the FOMC’s private information (discussed
in the text) were not in the public domain thereafter.
5

Like the oracle of Delphi, the forward guidance predicts the future but promises no specific policy actions.

6

For each FOMC meeting, we compare Board staff and Blue Chip forecasts for the current quarter and the
next four quarters. Since we also include forecast differences from the previous meeting in our model, this
would make for 30 coefficients to estimate. There were only 148 FOMC meetings during our sample period,
so including all of these variables would put us in danger of “overfitting” the model. To avoid this danger,
we reduced the forecast differences for each meeting and each variable to two variables using a statistical
procedure called factor analysis. We give the details of this procedure in Campbell et al. (2017).
7

7

See Campbell et al. (2017) for the details.

8

As a point of comparison, we showed in Campbell et al. (2017) that our measures of FOMC private information
have no statistically detectable influence on the current policy rate.
9

REFERENCES
Ben Zeev, Nadav, Christopher Gunn, and Hashmat Khan, 2017, “Monetary news shocks,” Carleton
Economic Papers, Carleton University, CEP 15-02, revised February 27, 2017 (originally issued March 2015).
Campbell, Jeffrey R., Charles L. Evans, Jonas D. M. Fisher, and Alejandro Justiniano, 2012,
“Macroeconomic effects of Federal Reserve forward guidance,” Brookings Papers on Economic Activity,
Spring, pp. 1–54.
Campbell, Jeffrey R., Jonas D. M. Fisher, Alejandro Justiniano, and Leonardo Melosi, 2017, “Forward
guidance and macroeconomic outcomes since the financial crisis,” in NBER Macroeconomics Annual 2016,
Martin Eichenbaum and Jonathan A. Parker (eds.), Vol. 31, Chicago: University of Chicago Press, forthcoming.

Economic Perspectives

3 / 2017

Federal Reserve Bank of Chicago

D’Amico, Stefania, and Thomas B. King, 2015, “What does anticipated monetary policy do?,” Federal
Reserve Bank of Chicago, working paper, No. 2015-10, November.
Eggertsson, Gauti B., and Michael Woodford, 2003, “The zero bound on interest rates and optimal monetary
policy,” Brookings Papers on Economic Activity, No. 1, pp. 139–233.
Krugman, Paul R., 1998, “It’s baaack: Japan’s slump and the return of the liquidity trap,” Brookings Papers
on Economic Activity, No. 2, pp. 137–205.
Jeffrey R. Campbell is a senior economist and research
advisor, Jonas D. M. Fisher is a vice president and the
director of macroeconomic research, Alejandro Justiniano
is a senior economist and research advisor, and Leonardo
Melosi is a senior economist in the Economic Research
Department at the Federal Reserve Bank of Chicago. In
addition, Campbell is affiliated with CentER, Tilburg
University; and Justiniano is affiliated with the Paris
School of Economics.
© 2017 Federal Reserve Bank of Chicago
Economic Perspectives is published by the Economic
Research Department of the Federal Reserve Bank of
Chicago. The views expressed are the authors’ and do not
necessarily reflect the views of the Federal Reserve Bank
of Chicago or the Federal Reserve System.
Charles L. Evans, President; Daniel G. Sullivan, Executive
Vice President and Director of Research; David Marshall,
Senior Vice President and Associate Director of Research;
Spencer Krane, Senior Vice President and Senior Research

Advisor; Daniel Aaronson, Vice President, microeconomic
policy research; Jonas D. M. Fisher, Vice President,
macroeconomic policy research; Robert Cox, Vice
President, markets team; Anna L. Paulson, Vice President,
finance team; William A. Testa, Vice President, regional
programs; Lisa Barrow, Senior Economist and Economics
Editor; Helen Koshy and Han Y. Choi, Editors; Julia Baker,
Production Editor; Sheila A. Mangler, Editorial Assistant.
Economic Perspectives articles may be reproduced in
whole or in part, provided the articles are not reproduced
or distributed for commercial gain and provided the
source is appropriately credited. Prior written permission
must be obtained for any other reproduction, distribution,
republication, or creation of derivative works of Economic
Perspectives articles. To request permission, please
contact Helen Koshy, senior editor, at 312-322-5830 or
email Helen.Koshy@chi.frb.org.
ISSN 0164-0682

8

Economic Perspectives

3 / 2017

Federal Reserve Bank of Chicago