Transcript “Communicating with Many Tongues: FOMC Speeches and U.S.

“Communicating with Many Tongues:
FOMC Speeches and U.S. Financial
Market Reaction”
by Bernd Hayo, Ali M. Kutan, and Matthias
Neuenkirch
Discussion by
Dennis W. Jansen
Texas A&M University
1
Recap
• Examines Speeches, Post-Meeting
Statements, Monetary Policy Reports,
Testimonies.
• Is there an impact on interest rates, stock
returns, and/or exchange rates?
– By types of communications (speeches, etc.)
– By entity doing the communicating
• Look at impact by BOG vs regional banks
• Or by chair of BOG, governors, regional bank
presidents, etc.
2
Recap of Results
• Impact on returns and volatility larger if
communication channel is ‘more formal.’
• Communications by BOG generally larger
impact than regional bank presidents
• Communications by chairman of BOG generally
larger than other BOG members.
• Some unexpected signs, lack of statistical
significance, indicate speeches may not by
themselves be that important for financial
markets.
3
Recap of Results
• To me, most intriguing results concern
news agencies. Authors report evidence
that news agencies perform a role in
filtering the information in speeches.
– Evidence is circumstantial, but related to prior
work by Ehrmann and Fratzscher (2007), who
use newswire information on Fed
communications
– Remember, Hayo and coauthors directly look
at the speeches, testimonies, etc.
4
Methodology
• Speeches (1439), congressional hearings (151),
post-meeting statements (68), and monetary
policy reports (20) were examined.
• Economic outlook can be ‘positive’ ‘neutral’ or
‘negative.’
• Monetary policy can be ‘tightening’ ‘neutral’ or
‘easing’.
• Difficulty: determining the surprise or news in a
speech.
– Can’t easily determine pre-speech expectations
5
Methodology
• Daily financial data Jan 1 1998 – Dec 29 2006
– Stock returns
– Foreign exchange returns
– Daily changes in rates on 3-mo, 6-mo, 1-yr, and 2-yr treasuries.
• Controls
– Foreign financial counterparts of the above, USD/Yen rate,
concensus forecasts of GDP, CPI, current GDP, CPI, GDP and
CPI 1 year ahead (why?)
– Surprise components: advance GDP, trade balance, I.P., ISM
manufacturing index and Conf Board consumer confidence
index, housing starts, nonfarm payroll, UE, retail sales, CPI, PPI,
all entered as impulse variables on the day of their
announcement.
6
Specification
• Returns = f(change in FFTR, consensus
forecasts, macro announcements, 9-11 dummy,
communications dummies) + h
• GARCH(1,1)
– h = h(macro announcements, communication events)
• There is a more inclusive model presented in the
paper, but it is not used because of estimation
difficulties.
7
Results – Some Anomalies
• Statement MP tightening at t-1 results in interest
rate declines (2 or 3 b.p.) at t. (Why?)
• Meanwhile, Statement MP tightening at t results
in a 3 b.p. increase in 2-yr rate at t.
– Why the discrepancy?
• Also: Speech MP easing at t+1 results in interest
rate declines (1-3 b.p.) at t for all but 2-yr rate.
8
Robustness Checks
• Authors compare their results with Ehrmann &
Fratzscher
– Differences due to differences in definitions, codings
– E&F use news agency events, not speeches
themselves.
– More news about Greenspan in the media than what
is in delivered official speeches.
• News agencies include information interviews, extra
information
• News agencies many not precisely identify who is speaking
on behalf of the FED in these informal matters. More may be
attributed wholly or partially to Greenspan than is warranted.
9
Intriguing Result
• Dating Greenspan speeches with the date they
appear in the media – not the date the speech
was given -- corrects the anomalous findings.
• Interp: Financial markets rely on media reports
delivered via news agencies. News agencies
act as filter. Financial markets react with a lag to
speeches even when those speeches are
delivered during market trading hours. News
agency reports tend to occur after hours or on
the next day.
10
More…
• News agencies do not cover all speeches
• They report on after-speech Q&A
• Apparently the news agencies play an
important information-transmission role in
the marketplace!
11
General Editorial Comments
• Discussion of estimation results is at times
a bit confusing. Authors present a model
that is not actually used for the relevant
results.
• The authors mention a novelty index, and
a repetition index. It is not clear to me
how these were constructed.
12
Suggestion re E&F
• Used E&F window, coding, etc., to
basically reproduce their results
• Might explore robustness of E&F results
– Asymmetric response to tightening/easing
– Asymmetric response to EO positive/negative
– Can you explore a longer window in the E&F
framework?
13
Suggestion re News Agency Results
• News agencies provide and filter Fed
communications.
• Is filtering less important when speeches or
testimony are covered by CNN or C-Span?
– Idea is such events are viewed in real time by traders.
– Put an indicator variable for when TV coverage was
available?
• Might allow estimation of impact of ‘filtering’ by
news agencies
• Of course, CNN (and C-SPAN?) coverage is not
random, involves selection issues by both CNN
and by the Fed itself.
14
Additional Comments
• Do all speeches that mention E.O. or M.P. convey
information over and above what is available in
measured financial variables?
– Related to issue of the surprise component of the speech.
– Communication about Fed actions or potential actions are most
clearly areas in which the Fed has superior information.
• How much do we care about the impact of a ‘typical
speech’ mentioning MP easing, or MP tightening?
• Perhaps we should look at specific speeches at specific
times.
• Tie results to theory?
– State of world at time of event should matter
– Exact nature of the communication should matter.
15
Additional Comments
• The fact that news agencies tend to filter
speeches also suggests that individuals might
likewise filter the information in speeches.
• News agency filtering of speeches is evidence
that not all speeches are alike.
• Maybe we should turn the problem around. We
can look at dates of individual communication
events (speech, or news article about speech)
and use the market response to indicate there
was (or may have been) important information in
the speech.
16
May 24 2001 Event: Speech by A.G.
EO: negative; MP: easing
Interest Rate Changes May 24 2001
0.25
0.20
0.15
0.10
0.05
0.00
-10
-9
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
9
10
-0.05
-0.10
-0.15
TB3M
TB6M
TN1Y
TN2Y
17
May 24 2001 Event: CAR’s
CAR May 24 2001
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
-10
-9
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
9
10
-0.05
TB3M
TB6M
TN1Y
TN2Y
18
April 21 2004 Event: Testimony by A.G.
EO: positive; MP: tightening.
19
April 21 2004 Event: CAR’s
CAR April 21 2004
0.15
0.1
0.05
0
-10
-9
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
9
10
-0.05
-0.1
TB3M
TB6M
TN1Y
TN2Y
20
11