DRAFT
Reactions of Canadian Interest Rates to Bank of
Canada Communications:
What moves Markets?
May 2008
Christine Fay1 and Toni Gravelle2
1
Financial Markets Department
Bank of Canada
Ottawa, Ontario, Canada K1A 0G9
cfay@bankofcanada.ca
2
Financial Markets Department
Bank of Canada
Ottawa, Ontario, Canada K1A 0G9
tgravelle@bankofcanada.ca
Bank of Canada working papers are theoretical or empirical works-in-progress on subjects in economics
and finance. The views expressed in this paper are those of the authors. No responsibility for them should
be attributed to the Bank of Canada.
ISSN ####-####
© 200? Bank of Canada
1
Acknowledgement
We wish to thank Scott Hendry, Christopher D’Souza, Rhys Mendes, Donna Howard,
Carolyn Wilkins, and Grahame Johnson for their helpful comments. We would also like
to thank Guy Mackenzie for his research assistance.
2
Reactions of Canadian Interest Rates to Bank of Canada
Communications: What moves Markets?
Abstract
We empirically examine the reaction of Canadian financial markets to official Bank of
Canada communication as well as their reaction to the recent use of “forward looking
statements” (a.k.a. monetary policy inclination statements) included in these
communications. This is done to investigate the extent to which Bank of Canada
monetary policy transparency has improved. We find evidence that Fixed Announcement
Date (FAD) press statements, and to a lesser extent speeches by Governing Council
members, significantly affect near-term interest rate expectations indicating that central
bank communication conveys important information to market participants. However,
our results also show that FAD press statements and speeches do not significantly impact
market rates over the more recent period when “forward looking statements” have been
used on a regular basis. We investigate two explanations for this change in response.
First, that market participants now better understand the Bank of Canada’s monetary
policy reaction function and, as such, there is less new incremental information conveyed
in these communications. Alternatively, that market participants have focused more on
the forward looking statements themselves and less on the Banks’ discussion of the
economic outlook (i.e., the Bank of Canada has become more predictable). As such the
participants would respond less to new macroeconomic data releases than before. We
find evidence to support the latter explanation.
3
1. Introduction
How do markets react to central bank communications? This question is of interest
because it helps assess central bank transparency, or more precisely, how well financial
market participants understand and use the guidance communicated by central banks to
effectively form their expectations for future inflation and asset market price movements
more generally. As such, this paper’s goals are twofold. First, we investigate the extent to
which Canadian interest rates respond to official Bank of Canada communication since
late 2000, the time since the Fixed Announcement Dates (FADs) were introduced, using
the same methodology as Reeves and Sawicki (2007). Specifically, based on a two stege
regression methodology, we test the impact of official communication on the variance of
interest rate using daily interest rate data. Following this, we study the impact of the Bank
of Canada’s use of forward looking statements on market behaviour. In effect, we want to
study whether the use of these statements have made the central bank more transparent
with regards to their reaction function or simply more predictable in terms of their future
policy rate actions.
As is now well documented in most central banking handbooks, transparency is
considered a key component of an effective monetary policy framework (see for example
ECB 2004). Central bank transparency makes monetary policy more effective in three
ways. First, in being clear about its mandate, how it goes about fulfilling it as well as its
ability and willingness to do so, the central bank engenders greater credibility. Secondly,
transparency, in regularly exposing the public to its views and understanding of current
and future economic activity, imposes some degree of accountability by providing the
means for the public to assess the consistency of the central bank’s actions and its
monetary policy makers’ decision making process with its stated objective. Thirdly, and
this is the main area of interest in this study, central bank transparency should help
markets understand the monetary policy reaction function, allowing markets to anticipate
the direction of future target interest rate movements. Monetary policy is more effective
if markets participants correctly anticipate it. That is, given that short-term rates and
long-term rates are linked via the expectations hypothesis, and the central bank only has
control over short-term (or more precisely overnight) interest rates, the central bank can
better influence longer-term interest rate expectations by communicating and shaping the
4
markets’ understanding of the factors driving the central banks reaction function and
economic outlook. A better understanding of the central bank’s reaction function, it is
thought, reduces market uncertainty about future rate movements (and implicitly market
volatility) and also allows a shortening of, and an increase in the effectiveness of, the
monetary transmission mechanism (the process by which expected changes in monetary
policy are incorporated into the movement of other financial variables and, in the end,
investment and consumption decisions).
Aside from transparency initiatives aimed at clarifying the Bank of Canada’s objective
(i.e., its inflation target) and monetary policy framework, the Bank of Canada has taken a
number of measures over the years to increase transparency and to communicate to the
public its views about the economic outlook. One significant step for enhancing
transparency and increasing the effectiveness of monetary policy was taken on October
30th 2000, when the Bank of Canada announced its first eight pre-specified “fixed
announcement dates” (FADs).
The introduction of FADs virtually eliminated the
market’s uncertainty regarding the exact timing of the Bank of Canada’s target rate
announcements, thereby reducing interest volatility related to this uncertainty. It also,
since the FAD schedule is known in advance, allowed market participant to focus more
attention on Canadian macroeconomic news that accumulated between FADs and
allowed them to speculate more precisely about the timing of target policy rate changes
(see Appendix B for more on this). In addition, and more important for this study, it also
gave the Bank an additional opportunity to increase the market’s information about the
Bank of Canada’s views on the economic outlook.
Previous empirical work undertaken at the Bank of Canada by Muller and Zelmer (1999),
Gravelle and Moessner (2002), Parent (2002-2003), and Andreou (2005) has studied
whether the move to FADs has increased central bank transparency. Parent (2002-2003),
extending the work of Gravelle and Moessner (2002) shows empirically that Canadian
market interest rates react to more Canadian macro announcement “surprises” and fewer
US macro announcement “surprises” in the post FAD period than in the pre-FAD period.
This work suggests that the introduction of the fixed announcement date regime has
increased transparency by shifting the focus of markets to Canadian rather than U.S.
economic conditions. As such it finds support for the idea that FADs have contributed
5
significantly to the improvement of the markets’ understanding of the Bank’s reaction
function. Andreou (2005), following closely the work of Kuttner (2000), measures
empirically the impact of policy surprises on Canadian government treasury bill and bond
yields. He finds that the impact of a surprise action on the long end of the yield curve has
diminished since the introduction of the FAD process suggesting that the Bank’s longterm policy goals are better understood and more credible.
Like most other industrialized central banks, the Bank of Canada has, over the years, also
increased its transparency by increasing the amount of information communicated to the
public. Specifically, the Bank has, since 1995, regularly published its Monetary Policy
Report (MPR) and update (MPRU). It has also, for sometime, published a press statement
with changes in the policy rate. For many years speeches by Governing Council (the
Governor and five Deputy Governors) have provided an opportunity to provide monetary
policy information to the public. But like the press statements that accompany policy
decisions, their potential to impact market expectations of future interest rates, were
enhanced significantly with the introduction of FADs. As such, our study focuses on
Bank of Canada communications for the period since FADs were introduced.
Another more recent measure the Bank of Canada has taken to increase transparency is
the inclusion of “forward looking statements” in FAD press releases and MPRs (and
MPRUs). In the following section, we discuss in detail how the publication of policy rate
guidance or a policy interest rate path might make a central bank more “predictable,” but
does not necessarily increase the market’s understanding of the central bank’s monetary
policy reaction function and as such does not necessarily make it more “transparent.”
Any increase in transparency hindges on the central bank’s ability to sufficiently convey
to market participants the conditionality embedded in the forward looking signal about
the future policy actions or policy rate path.
Empirical work measuring the impact of central bank communication is now quite
extensive. Comparing the communication strategies of the Federal Reserve, the Bank of
England and the ECB, Ehrmann & Fratzscher (2005 and 2007a) find that monetary policy
communication generally has a significant effect on the short and medium-term horizons
6
of the yield curve.1 Using daily data, Kohn & Sack (2003) find that FOMC statements
accompanying policy decisions and congressional testimony given by Chairman
Greenspan have a significant effect on interest rates. Following the methodology of Kohn
& Sack (2003), Reeves & Sawicki (2005) study the impact of official Bank of England
(BoE) communications and find that only the publication of the Minutes of the Monetary
Policy Committee meetings significantly impact market volatility when using daily data.
However, when using intraday data, they find that both the Minutes and the Inflation
Report have a significant impact on near-term interest rate expectations. Looking at a
panel of six central banks including the Bank of Canada2, Connolly and Kohler (2007)
find that across all countries, commentaries accompanying rate decisions have the largest
impact. Among Bank of Canada communications, press releases accompanying FAD
decisions, MPRs and MPRUs3, and speeches were the forms of official communication
that they found to have a statistically significant impact on interest rates4. The results
differ from ours in that we focus on the post-FAD period, while the Connolly and Kohler
examine the January 1997 to August 2006 period. They also found that across countries,
central bank communication had a small explanatory power for movements in interest
rates overall relative to movement of global interests which is proxied by benchmark U.S
rates.
We find evidence that official Bank of Canada communication, in particular FAD press
statements, have a significant effect on near- to medium-term yields suggesting that these
communications convey important new information which impact the markets interest
rate expectations. As mentioned, over our period of study there is a major shift in the
Bank of Canada’s communication strategy that saw the inclusion of forward looking
statements in FAD press statements and MPRs, which allows us to study the impact of
these statements on Central Bank transparency. Reruning our tests on both a split sample
(where these statements began being used in a consistent manner) as well as cross1
They also find that statements about the economic outlook only have a significant impact on the medium
to long end of the yield curve in the US. They suggest this finding is related to differences in the stated
monetary policy objectives of the three central banks. The Bank of England and the ECB focus on price
stability whereas the Federal Reserve gives a stronger weight to the real economy.
2
This panel includes the central banks of Australia, Canada, the Euro Area, New Zealand, the UK and the
US.
3
When Connolly & Kohler re-estimated their model over the post-FAD period (2000 – 2006), MPR(U)’s
no longer had a significant impact on market rates.
4
Market rates used for Canada in this study included the front 8 BAX futures contracts.
7
dummies to account for dates prior to our split that include FLS’s, our results suggest that
there has been a loss of conditionality surrounding Bank of Canada communications with
the introduction of these forward looking statements, which is indicative of greater
“predictability” but not of greater transparency measured in terms of the market having a
better understanding of the monetary policy reaction function.
2. Issues Related with Publishing Policy Interest Rate Guidance or Paths:
Predictability versus Transparency and Communicating Conditionality
There are different aspects of monetary policy transparency. In this study we focus on
one aspect of transparency by defining monetary policy as being transparent if market
participants can anticipate the central banks’ official interest rate decisions correctly,
based on their knowledge of the central banks’ reaction function. This definition is
motivated by the view that central bank communication should be used to increase the
effectiveness of monetary policy as described in the introduction.
Under this definition then, in principal, central bank communication is put forward to
help market participants update their understanding the central bank’s reaction function
parameters, or more generally, the framework the central bank is currently relying upon
to guide their decision making process. As such, communications about the central
bank’s current view of how the economy is unfolding, typically summarised in
projections of output or inflation that include, in some situations, some form of
uncertainty bands (i.e., fan charts) help in this regard.
In addition, communication
regarding the macroeconomic variables that are particularly in focus in formulating
policy, that summarizes how the central banks see certain alternative scenarios unfolding
for the outlook, and that presents updates to the model or the modeling process, should
enhance the markets understanding of the central bank’s reaction function and also cause
a reaction in interest rates, both at the short end of the yield curve as well as in medium-,
and in some cases, longer-term interest rates.
Recently, a debate has unfolded around how much additional information central banks
should release to the public with respect to their future policy rate intentions.
8
In
particular, should central banks provide to the public their forecast of, or their intention
for, the path of future target interest rates? Central banks see both pros and cons in
providing policy rate inclination signals (including the publishing of the policy rate
path).5 (Note that we focus only on the pros and cons as they relate to increasing the
effectiveness of the monetary policy transmission mechanism, while Kahn (2007) and
others also include those related to increasing monetary policy accountability and
credibility.) One of the main advantages of providing policy rate guidance is that, in
principal, it makes monetary policy, via the expectation hypothesis, more effective (i.e.,
more bang-for-buck)6 by better influencing medium- and long-term rates as these are
likely to react more to policy actions that are accompanied by communication about the
path of future policy rates. For example, if a central bank communicates that it will hold
policy rates higher for a longer period of time than markets currently expect, then
medium- and long-term rates would likely be higher than if the central bank simply
communicated the current target for the policy rate (Kahn 2007)7. Another advantage of
providing guidance is it reduces the degree of market uncertainty related to future
monetary policy moves and thus reduces interest rate risk premia.
Rudebush (2008) characterises policy inclination signals as falling into three categories:
indirect signals; direct qualitative signals; and direct quantitative signals. The first,
“indirect signals” provides implicit information about the policy path through the use of
related information. Examples of indirect signals include balance of risk statements, the
presentation of a risk scenario showing the extent that inflation would deviate from the
target holding policy rates constant.
According to Rudebusch (2008), examples of the second type of policy guidance, “direct
qualitative” signals, include the policy “bias” statements used for a period by the Fed
starting in the late 1990s, as well as phrases that signal the desired policy stance over an
extended number of meeting dates, such as policy accommodation “can be maintained for
a considerable period” or “can be removed at pace that is likely to be measured” also fall
5
See Kahn (2007) for a summary as well as Moessner and Nelson (2008).
However, this hypothesis, to our knowledge, has not been directly tested empirically.
7
There is a third advantage put forward by theorists. When private agents are uncertain about the central
bank’s inflation objective (particularly those central banks with new objectives or targeting regimes), it
helps, in theory, align the private agents’ and the central bank’s expectations about future monetary policy
and facilitating the economy’s adjustment back to the inflation objective after a shock.
6
9
in this category according to Rudebusch (2008). The final category, “direct quantitative”
signals, best describe the explicit numerical projections provided by the central banks of
New Zealand, Norway, Iceland, and Sweden
The ECB’s use of “code words” such as “strong vigilance” would seem to fall into the
indirect qualitative signal category. The Bank of Canada has provided direct qualitative
signals to markets via its forward looking statements included in nearly all FAD (and/or
MPR) press releases, since July 2004. These have typically included statements such as
“some increase in the target for the overnight rate may be required in the medium (near)
term” or “the current level of the target for the overnight rate is consistent with achieving
the inflation target over the medium (near) term” or “… further reduction of monetary
stimulus will be required ... over the next four to six quarters…” (see Table C.1 in
Appendix C). The Bank of Canada also has recently introduced “balance of risk to the
outlook statements” which would be categorized as “indirect guidance”. Moreover, the
Bank has provided both indirect and direct qualitative signals in its other
communications, particularly the MPRs and speeches that contained language consistent
with the MPRs.
Kohn (2005) and others, have highlighted that there are notable disadvantages to
providing guidance. A major disadvantage of providing policy rate guidance is that
markets might place too great a weight on the central banks guidance and as a
consequence not fully understand or appreciate sufficiently the conditionality of this
forecast or guidance. This can result in markets focusing less on their own private
information (won’t do their homework). It may also reduce the information content of
market prices (for information extraction purposes). A second disadvantage related to the
perceived un-conditionality of the policy rate guidance is that it might cause policy
makers to be less willing to change their policy intentions in light of new information,
due to the following two concerns. First, frequent updating of the policy path might
undermine the public’s confidence in the central bank’s forecasting ability. Second,
policy markers may be concerned that financial markets will overreact to a shift in policy
stance or guidance, leading to excess volatility. In other words, policy makers might be
concerned that by “disappointing expectations” they will engender market volatility, even
10
though the change in circumstance justifies their reassessment of the appropriate policy
action.
A third problem with providing policy rate guidance is that it may in itself be difficult to
provide forward guidance given the structure of the monetary policy decision making
process and/or committee. For example, given its governance structure, the Bank of
England’s Monetary Policy Committee’s basis for its decision is “renewed or refreshed”
every month and as such is “very” conditional on the new information they’ve received
since their last meeting (i.e., they don’t decide in advance what their decision will be at
the next (few) MPC meeting(s) but instead meet with a relatively clean slate every time).
They are thus averse to potentially fooling financial markets, via their communications,
into thinking there are definite multi-meeting-date plans for the policy rate going
forward. King (2006) notes that the added benefit of not providing forward guidance, is
that there is no need to worry about how to wean markets off their “crutch of spoon fed
expectations formation” without causing unintended volatility, when the central bank
sees much greater uncertainty about the future path it will take, and is no longer able to
provide relatively unconditional policy inclination statements.8
In considering the issue of providing policy guidance in official central bank
communication, it is important to note that there is a subtle difference in a central bank’s
communication strategy that is “transparent” and one that is “predictable” (see Moessner,
Gravelle, and Sinclair (2005) and Jen 2007). Conceptually, a more “predictable” central
bank can be thought of one that provides more policy “guidance” via, for example,
“policy inclination statements” or their forecast/projection for future path of the policy
rate, which explicitly includes information on the direction and perhaps timing of future
interest rate changes. A more predictable central bank is one for which markets
participants can more easily anticipate correctly the next policy decision (or the next set
of policy decisions) without necessarily a better understanding of the reasons for the
decision(s) and, as such, without a better understanding of the central bank’s reaction
8
This is roughly what happened to the FED at the April FOMC meeting, in which the press release no
longer made it explicit that the FED will continue on this path.
11
function. On the other hand, a more transparent central bank is one that effectively
conveys to the market their monetary policy reaction function (as described above).9
Although the aim of policy makers in providing policy guidance may be, in general, to
enhance the markets’ understanding of the reaction function and ultimately the
effectiveness of monetary policy, because of the market participants’ perverse (from the
policymakers perspective at least) focus on the guidance communicated as described
above, it turns out to reduce the markets’ reliance on their own private information,
reduces their reaction to macro news, and reduces their incentives to update their
understanding of the reaction function. As such, greater predictability might not
necessarily imply greater monetary policy transparency, but greater transparency (in the
sense that communication of information that effectively enhances the markets’
understanding of the reaction function) does in general imply greater predictability.
However, it is not clear that those central banks that publish their target rate paths or
some other form of policy rate guidance are necessarily “predictable”, as predictability is
dependent on the degree of perceived conditionality (or in particular the lack thereof)
embedded in the central banks communications, including in any one of the three types of
policy guidance. It is possible that central banks that provide a direct quantitative
guidance (i.e., that publish future target rate paths) can be less predictable (and more
transparent) than those that simply offering direct qualitative guidance (i.e., that publish
policy inclination statements), if the inclination statement is explicitly presented to, or is
implicitly perceived by, financial markets as less conditional than the policy path.
Specifically, central banks could provide “error,” “risks,” or “uncertainty” bands around
their forecast of key economic variables, including the policy rate, and explicitly indicate
9
Strictly trying to increasing the markets’ understanding of the monetary policy reaction function (i.e.,
increasing transparency as defined above) on the other hand does not lead to the same problems of
perceived un-conditionality and yet still allows for enhanced monetary policy effectiveness. If markets had
a perfect understanding of the reaction function, they could correctly anticipate the path of future policy
rates to, in principal, the same degree as if the policy guidance was provided. By not stating explicitly the
implications of its economic outlook for the future stance of monetary policy, a central bank would force
market participants to draw their own conclusions for future policy stances. Market participants could do
this by focusing on the implications of economic and financial developments, rather than by simply
speculating on whether and when the Bank will follow through with its previously announced “guidance”
for monetary conditions.
12
that the timing of policy changes could occur anytime over a lengthy period (i.e., anytime
within the projection period). Moreover, central banks that publish a target rate path
could in principal use this as a tool to aid their (conditionality laden) communication
about the economic outlook, providing in-depth alternative scenarios and/or risks to their
base case policy rate projections. Central banks could indicate that the path is simply the
mean/mode of a series of, or probability distribution of, scenarios.
In addition, Moessner and Nelson (2008) argue that the regular appearance of a forecast
policy rate path in central bank communications may in itself make these more
conditional in nature relative to those central banks that irregularly communicate
guidance, as the latter may be viewed as doing so for the tactical reason of “massaging”
market expectations. The latter central banks’ communication guidance strategy may
thus look more like unconditional commitments in this case. Nonetheless, it would seem
that central banks that provide direct qualitative or quantitative guidance in the form of a
forward looking statement or publishing a policy rate path, have more “work” to do in
terms of making clear the high degree of conditionally embedded in their
communications.
In sum, what is important in terms of measuring central bank predictability is one: to
what extent the central bank conveys the timing and direction of future rate changes; and
two: how much conditionality there is explicitly embedded in, or implicitly perceived by
the market in, its communications. As highlighted by Kahn (2007), central banks that
restrict themselves to use only “balance of risks” statements leave “the markets to
interpret any possible implication of these risks for the (future) policy rates.” In contrast,
policy statements like the forward looking statements used by the Bank of Canada or the
“measured paced” guidance provided by the Fed over an extended period of time, seem
more unconditional in nature.
Empirically, central banks that are increasingly predictable should see a decrease in the
financial markets reliance on macroeconomic news to anticipate near-term monetary
policy changes. The relatively unconditional nature of the forward guidance provided,
would negate the need for markets to do their homework in terms of formulating their
“shadow” monetary reaction functions and in turn using this to calculate the timing of the
13
next, or the next few, target rate changes. As mentioned in the introduction, this is in fact
what Ehrmann and Fratzscher (2007b) find when studying changes in U.S. Federal
Reserve communications. They find that since the Fed introduced its policy inclination
statement in 1999, interest rate reactions to macro news were generally smaller and there
were fewer macro releases that were statistically significant. Moreover, they find that
inter-meeting communications about the future course/timing of monetary policy, rather
than that about the economic outlook, garnered smaller interest rate reactions after the
Fed introduced its policy inclination statement. In this study, we investigate whether or
not the inclusion of forward looking statements in Bank of Canada communication has
caused markets to react less to macroeconomic releases.
3. Types of Central Bank Communications
In our paper we focus on three main types of official Bank of Canada communications
These include: Fixed announcement date (FAD) press statements; the Monetary Policy
Report and Update; and, Speeches and other communications comprising press
conferences and interviews. Given that FAD press releases are one of the communication
items, we begin our sample on October 30, 2000. This is the date the Bank of Canada
published the first schedule of FADs, making it the first date on which market
participants could trade based on this knowledge. Our sample ends on May 31, 2007.
3.1. FAD Press Statements
On September 19, 2000 in an effort to increase transparency, the Bank of Canada
announced that it would move to a FAD regime, changing (or not) its’ target for the
overnight policy rate on eight pre-specified dates each year (except in exceptional
circumstances). The first of these FADs occurred on December 5, 2000. On each fixed
announcement date, a press statement is released at 9:00 am (eastern time) announcing
the policy decision. The statement includes the reasons underlying the policy decision,
an update of the governing council’s view of the economic outlook, and, more recently,
forward looking policy guidance and a discussion about the balance of risks to the
outlook. As such, it is an important form of communication and receives intense scrutiny
by market participants.
14
Between October 30, 2000 and May 31, 2007 ( our sample period) there were 53 FADs
with accompanying press releases. There was also one intra-meeting policy change with a
press release on September 17, 2001, shortly after the September 11, 2001 terrorist
attacks. For most of this paper we do not include the period following and including
September 11, 2001 (3 month window) as this period is not representative of the normal
market dynamics or normal reactions to communication.
3.2 Monetary Policy Reports (MPRs) and MPR Updates (MPRUs)
The MPR and its update are released four times a year, two days following every second
FAD at 10:30 am (Eastern Time).10 The Governing Council (GC) uses this document as
its main method of communicating and updating its detailed views on the current state
and likely evolution of the economy. Although the main messages of the MPR(U) have
been summarized in the FAD press statement, the MPR(U) elaborates in greater detail the
Governing Council’s assessment of the factors that have shaped the interest rate decision,
and puts into context recent developments in terms of the underlying trends over the
medium term.
Prior to the release of the MPR, the GC, with substantial input from the Bank of Canada
staff, comes to a consensus view on the future path of the Canadian economy, taking into
account any new information since the previous MPR. It is therefore the main vehicle for
the GC to communicate in detail, any significant changes in its view of the economic
outlook. As such, it plays a role akin to some central banks’ “minutes” in discussing the
factors underlying the Bank of Canada’s economic and inflation outlook. However, given
Governing Council’s monetary policy decision making process is consensus-based, the
MPR (as opposed to certain central bank minutes) does not provide individual GC
member views on the outlook.
10
Since the inception of Fixed announcement dates, the MPR(U) release has always followed the FAD
press release. The time lag between these two releases has shortened over time (although the content and
consistency between the two has not changed materially). The MPR(U) was initially released two weeks
following the FAD. In mid-2001 the Bank began releasing the MPR(U) within the same week as the FAD,
and currently, the MPR(U) is released two days following the FAD.
15
Our sample contains 14 MPRs and 13 MPRUs for a total of 2711.
3.3. Speeches
All governing council members give speeches and interviews. The views expressed in
BoC speeches are representative of, and consistent across, all governing council
members, rather than the views of individual members. This is a function of the fact that
the underlying monetary policy decision structure at the BoC is by consensus.
Governing council members speak on a number of topics, which may or may not be of
interest to market participants. However, many speeches contain an economic
conjuncture as well as a question and answer period following the speech. In general, the
economic outlook section in a speech is used to reinforce and elaborate on the key
messages in the MPR(U) and FAD press statement, and it is only when there are
developments that lead governing council to significantly change their views from what
was most recently laid out, that they would seek to provide a clear signal of this change in
view. In addition, there is a black out period the week preceding the FAD in which the
BoC does not comment on either the economy or the direction of monetary policy.
In order to include only those speeches relevant to monetary policy or the economic
outlook, as well as capture information that the market actually receives, we include only
those speeches, interviews and news conferences that generate at least one story headline
in Bloomberg related to either monetary policy or the economic outlook. Over our sample
period GC members made 153 speeches, of which we include 98.
3.4 Forward Looking Statements (FLSs) and Balance of Risks Discussions
The Bank of Canada began using FLSs in April 2002, where they appeared in both FAD
press statements as well as MPRs.12 See Table 1 in Appendix B which outlines the
11
The main difference between the MPR and MPRU is it’s length and the depth of analysis of factors
outside the domestic economy. We do not believe that this difference has a significant impact on market
reaction. As such, we will treat the MPR and MPRU as the same event.
12
Forward looking policy statements began appearing in the highlights section of some of the MPR(U)’s
and their accompanying press release with the April 2002 MPR and were included in all MPR(U)’s
beginning with the July 22, 2004 MPRU.
16
wording of each FLS. It was not until July 2004 that a FLS was issued with nearly each
FAD and/or MPR. It is important to note that over our sample, there have only been
either positive or neutral FLSs. There are two reasons for this. First, the period since July
2004 where the Bank of Canada has more consistently included FLSs, had also been a
period where there has been a sustained economic growth period and as such where the
policy rate was on a relatively consistent upward march (see Chart 1). It is also worth
noting that in general, subsequent monetary policy actions have, in general, been
consistent with these statements as outlined in Table 2, Appendix B.
There also have been the more qualitative “balance of risk” statements issued with FAD
and MPR releases. Although we do not examine these specifically in our study, we
thought it useful to provide a brief summary of their characteristics. Although a few
balance of risk statements were issued earlier in the FAD period, the consistent inclusion
of these in all FAD/MPR press releases started in July of 2005. These statements are
presented in Table 1 in the Appendix B. Note that the earlier set of balance of risk
statements were split between short-term (or near-term) risks to the projection and those
“later on” in the projection (or over the medium term). The near-term risks were always
balanced and those later on were generally tilted downwards. Starting in September
2006, the Bank no longer split the timing of the risks and simply referred, in general, to
“the risk to its inflation projection”. After several press releases which stated risk were
balanced, the first balance of risk statement to indicate unbalanced risks over the
projection period occurred in April 2007, when Bank said “risks are roughly balanced,
but with a slight tilt to the upside.” However, this “composite” balance of risk statement
might not provide a very clear signal of the Bank’s view of the balance of risk. The first
unequivocal message occurred in December 2007, roughly two and half years after the
balance of risk statements began to be consistently included in the press statements, when
the Bank saw “a shift to the down side in the balance of risks around … the projection for
inflation through 2009.”
4. Methodology
4.1 Measuring Market Reaction to Communications
17
Four problems arise when trying to measure the market impact of communications. The
first is that it is difficult to quantify and characterise systematically the content of Central
Bank communications, making it hard to benchmark the “strength/importance” or
“direction” of what is announced. Specifically, it would be useful to quantify the
direction or implied stance of all that is communicated by the Bank, by for example
categorizing it into dovish, neutral, or hawkish statements. But this is at best a possibility
only for a subset of FAD/MPR press releases that included a forward looking statement,
rather than the complete set of communications. Secondly, we cannot easily measure
what markets had expected these communications to say, therefore adding to the
difficulty in assessing the strength or sign of the signal.
To address these two problems we follow the methodology of Kohn & Sack (2003) and
Reeves & Sawicki (2005) in which we focus on the volatility of interest rates on days of
official Bank of Canada Communication. Since any change in the mean should also effect
the variance of the asset price, this is a good proxy for analysing the overall importance
of the communication event for financial markets without needing to quantify the
strength or direction of the event itself. Thus, we would expect to see the volatility of
these instruments rise on communication days if this communication is passing along
new relevant information as this news is incorporated into market rates.
Another issue when measuring market reaction to communications is that the content of
communications or the communication event itself may be endogenous, that is, the
central bank may choose to communicate new information, or increase the frequency of
its communication because of a sudden change in the economic conjuncture or some
other news. In this case, asset prices would probably be more volatile on the days of
communication, possibly in reaction to other factors we are unable to control for, thus
possibly overstating the impact of the communication. With respect to the endogeneity of
communication event, outside of speeches, the Bank of Canada sets and announces its
communication dates well in advance, therefore ruling this out to a large extent as an
issue in our study.
The final complication that occurs when attempting to measure the impact of
communications on interest rates is that other events taking place on the same day as
18
these communications may also be relevant to market participants and therefore move
interest rates. These factors include macroeconomic news in both Canada and the U.S.,
U.S. and Canadian policy rate surprises, and U.S. monetary policy communications. We
control for U.S. news variables due to the close economic links between Canada and the
U.S. and the fact that that changes in U.S. benchmark interest rates tend to move other
industrialized country rates. As such, in order to isolate the asset price movements linked
to official Bank communications, we must first take into account all other relevant
information that could move interest rates on communications days. Our analysis
therefore takes on two distinct parts.
4.2 Regression equations
To control for the effect of macroeconomic announcements, policy surprises and U.S.
communications on asset prices, we estimate equation (1) using OLS13, following Kohn
& Sack (2003), and Ehrmann & Fratzscher (2005) among others, with modifications for
the Canadian economy. The first stage regression model is as follows:
n
m
i =1
j =1
Δy t = β 0 + β 1 ΔON tu + β 2 Δff t u + β 3 Δef t + β 4 ΔT 2 t + ∑ α i cmaciu,t + ∑ α j usmaciu,t + ε t
(1)
where the one-day change in the interest rate of interest, Δy , are regressed on the surprise
component of Canadian policy announcements, ΔON tu ; the surprise component of U.S.
policy announcements, Δff u ; FOMC communication control variables, Δef and ΔT 2 ;
and the surprise component of macroeconomic announcements in Canada and the U.S.;
cmacu and usmacu respectively. We then relate the unexplained variance of market prices
from this regression to communications variables as described below in Section 4.3.
Please see Appendix A, where we describe in detail each of these controls.
4.3 Measuring the Impact of Communications on Market Rates
13
We use the Newey-West estimator which “provides a way to calculate consistent covariance matrices in
the presence of both serial correlation and heteroscedasticity.” (Johnson & Dinardo,1997, p.333).
19
Once we have controlled for “other news”, we measure the impact of official Bank
communications on markets by relating the unexplained variance to these
communications using the following equation.
ε
2
i ,t
n
= δ 0 + δ 2Vixt + ∑ γ j comm j ,t + η i ,t
(2)
j =1
Where ε t2 is the squared residual of from equation (1) for interest rate i. Commj,t
represents the jth type of communication, which in our study are the FAD press releases,
MPR releases, or speeches. We will set this variable equal 1 on communication days and
0 otherwise. We will compare the variance of these market rates on communication
against the average variance on all non-communication days controlling for the gradual
decline in market volatility over our period of study by including the VIX index.14
Specifically, if Bank communications in fact convey important ``market moving``
information then we expect that γj,t to be positive and statistically significant.
Section 5: Data
Our data consists of daily observations over the post-FAD period: October 30, 2000 to
May 31, 2007.15 We exclude the three months following the September 11, 2001 terrorist
attacks, in line with the methodology of Kohn & Sack, due to the disruption in financial
markets around that period that may lead to distortions. This makes our sample size 2313
days for each regression.
In order to study the impact of communications on both short as well as longer term
interest rate expectations we use yields of different maturities as the dependent variables.
At the short end we use the 3 month CDOR16 interest rate as this is the rate to which the
14
Kohn & Sack (2003) argue that a comparison of the volatility on FOMC statement days against the level
of volatility observed over the week preceding each statement is a superior measure as it better controls for
patterns of volatility over the sample. We find however, that including a dummy for the week prior to a
communication event has little impact on our results. As well, we find, as in Kohn & Sack, and Reeves and
Sawicki, that estimating equation (5) using GARCH (without including the VIX) does not change our
results.
15
October 30, 2000 is the date the Bank of Canada issued a press release stating the 2000-2001 dates for
the FAD
16
The CDOR rate is tabulated by Reuters at 10:00 and is a survey of the 9 major Canadian dealers. The top
and bottom rates are dropped and the rest are averaged. The CDOR is announced at 10:25.
20
BAX futures contracts settle and was found by Johnson (2003) to be a good measure of
market expectations. As well, we include the front BAX contracts17 and 2-, 5-, & 10-year
Government of Canada benchmark bond yields; however, we first needed to make an
adjustment to these yields as there were issues with duration and rollover with both
series. That is, for these series, the actual duration of the instrument changes on a daily
basis, with big jumps when the contract changes or rolls over. We therefore calculated a
90 day, 180 day and 270 day constant maturity futures-based interest rate by linearly
interpolating between the rates on the front four BAX futures contracts. We will refer to
these contracts as BAX1, BAX2, and BAX3 respectively. Next, consistent with Reeves
and Sawicki, we calculated 2, 5 and 10 year constant maturity bond yields using the zero
coupon curve, available on the Bank of Canada website. These yields were used in a
number of Canadian studies looking at market reaction including Gravelle & Moessner
(2002) and Parent (2002-2003). For a more detail description of the control variables
including the macroeconomic surprise variables see Appendix A
5.2 Macroeconomic announcements data
Using OLS to estimate equation (1), we include in our study the subset of independent
macro surprise variables that were significant at the 5 per cent levels over our sample (see
appendix C). Among For Canada, this includes both the core and headline consumer
price index releases, the employment & GDP releases, housing starts, the Ivey purchasing
managers index, leading indicator, manufacturing shipments and retail sales. The U.S.
macro surprise variables we find significant include, core CPI, GDP, hourly earnings,
industrial production, ISM, non-farm payrolls, the core and headline producer price
indices, the trade balance and the unemployment release18.
6. Results
6.1 Market Impact of Official Bank of Canada Communications
17
Johnson (2003) shows empirically that the front three BAX contracts are among the rates that are most
representative of expectations in Canada (under 1 year) and Harvey (1996) shows that changes in futures
prices tend to respond more quickly than (or lead) other money market rates in their reaction to economic
news.
18
The Ivey Purchasing managers index begins in January 2001
21
As can be seen in Table 1, FAD press statements and speeches have a significant impact
on the volatility of market rates over the full sample, October 2000 to May 2007. The
FAD press statements are significant for all BAX interest rates and the 2-year yield and
have the largest impact in the 2nd and 3rd BAX. This implies that market participants, on
average over the sample, find that press releases provide important information for the
short to medium term outlook. Yields longer than 2 years are not significantly impacted
by any type of communication.
Interestingly, the MPR was not significant in any of our tests. This could be due to one of
two things. First, in the post-FAD period, MPR(U)s closely follow FAD press statements.
It is possible, then, that these FAD press statements are “scooping” the new information
in the MPR. As discussed earlier, Connolly & Kohler (2006) found that the MPR was
significant, however, their sample covers the 1997 to 2004 period. When they run their
test over the shorter post FAD-only period, they find that the MPR is no longer
significant. This is consistent with the fact that in the pre-FAD period, the MPR did not
follow the rate decision announcement by any pre-specified time period, as these
decisions were made on an as needed basis.
Another possible explanation is that there is an impact but as we are using daily data and
are unable to control sufficiently for other news that day, this other news is drowning out
the impact of the communication event. This theory is supported by the findings of
Reeves and Sawicki who found that the BoE Inflation Report was not significant at a
daily frequency, but significant when they used intraday data.
Table 1: Full sample results
n
ε i2,t = δ 0 + δ 2Vixt + ∑ γ j comm j ,t + η i ,t
j =1
Increase in Var(ε) due to:
Interest Rates Var(ε) noncomm days
3.116
90D CDOR
(0.000)
13.367
BAX1
(0.000)
Fad Press
Release
0.772
(0.392)
14.761
(0.001)
22
MPR(U)
Speeches
5.944
(0.289)
17.014
(0.295)
0.261
(0.776)
6.431
(0.060)
BAX2
BAX3
2Y Bond`
5Y Bond
10Y Bond
23.701
(0.000)
28.621
(0.000)
17.882
(0.000)
16.990
(0.000)
14.634
(0.000)
24.463
(0.212)
23.963
(0.234)
14.333
(0.249)
4.547
(0.519)
-0.833
(0.863)
24.930
(0.004)
26.570
(0.010)
14.975
(0.023)
5.146
(0.283)
-0.251
(0.934)
15.586
(0.021)
16.241
(0.037)
5.236
(0.177)
2.121
(0.460)
0.432
(0.844)
Given that speeches rarely, as discussed above, deviate from the discussion presented in
the published MPR, the finding that speeches had an impact on markets, over the full
sample, while the MPR did not, was a bit of a puzzle at first. However, as can be seen in
Graph 1 below, which displays our volatility measure (the squared residuals from the
estimated equation 1) for the BAX2 rate over our sample (the red bars) we see that there
are two clusters of high volatility days that help explain this result (note that the green
line in this graph is the average volatility on non-communication days). Both of these
high volatility periods are marked by higher uncertainty in markets due to either financial
headwinds from accounting scandals such as Enron and Worldcom in the earlier period
or SARs, the East Coast Blackout and a rapidly appreciating Canadian Dollar in the
second period. These periods are also significant in that a number of speeches included
inter FAD updates to the Bank’s views about the future direction of monetary policy
(which would cause significant volatility in market prices as “new” information is
incorporated into market prices), and that during these periods FLSs were not used in
FAD press statements or MPR(U)’s.
To test this hypothesis empirically, we conducted a sensitivity analysis by removing one
by one, the speeches that had the largest impact on our market rates. By removing only
two speeches (2 per cent of our sample), our results were no longer significant at the 5
per cent level, thus providing support to our theory that it is only a handful of speeches
that impact market rates.
23
Graph 1
400
7
350
6
Enron
Worldcom
Financial Headwinds
300
250
5
SARS
East Coast Blackout
Strong appreciation of
Canadian Dollar
4
200
ON Rate Target
3
150
100
2
Var BAX2 on nonCommunication
Days: 23.7
BAX2
50
0
O
ct
-0
0
M
ar
-0
1
Ju
l-0
1
N
ov
-0
1
M
ar
-0
2
Ju
l-0
2
N
ov
-0
2
A
pr
-0
3
A
ug
-0
3
D
ec
-0
3
A
pr
-0
4
A
ug
-0
4
D
ec
-0
4
A
pr
-0
5
A
ug
-0
5
Ja
n06
M
ay
-0
6
Se
p06
Ja
n07
M
ay
-0
7
0
1
7. Impact of forward looking Statements on Asset Prices: Are Markets viewing
these as unconditional commitments
Johnson, Johnson & Fay (2006) define FLSs as communications that explicitly state
Governing council’s view of the likely future path for policy rates. In this section we are
interested in how these statements have impacted market behaviour. More specifically,
we examine empirically whether these statements have helped markets in their
understanding of the Bank of Canada’s monetary policy reaction function or instead have
only made the monetary policy actions of the central bank more predictable. If the latter
is true, this would cause market participants to react less to information that is relevant in
calculating the reaction function in the near term, and in turn reduce the information
content embodied in rates.
As previously mentioned, some argue that FLSs improve transparency about the future
direction of monetary policy and as such could reduce volatility in interest rates and
possibly increase the central banks’ influence over the full range of the yield curve,
potentially enhancing the effectiveness of monetary policy actions. On the flip side of the
24
debate, the main concerns are that this policy guidance could be taken as a precommitment as it is difficult to condition the statements, and that it may lead to less
independent analysis of the economy by the market.
If market participants do in fact view FLSs as a rough pre-commitment and the consistent
presence of a FLS makes the Bank of Canada predictable, then outside of the information
provided in unanticipated changes in the FLS, the information about the Bank’s view of
the economic outlook would be less in focus and thus longer-dated BAX rates and bond
yields should move less on FAD days. In addition, market participants should react less
to macroeconomic news given the consistent inclusion of FLSs.
We first examine the hypothesis that interest rates will react less during the period when
the FLS is included on a regular basis in the FAD and MPR statements. This is done in
two ways. We first split our sample in two at July 22, 2004, the point at which the BoC
began using FLSs in a consistent manner, and examine the communication coefficient
estimates over this July 2004 to May 2007 sample period.
Tables 2a and 2b provides estimates of the communications coefficients over the first and
second sample respectively. In the first sample, a period where the FLS was used
inconsistently and sparingly, the FAD and speeches are significant for various maturities
of interest rates. In Table 2b we can see that there are, for all but one interest rate, no
longer any communication events that are significant.19
Table 2a: First Subsample Results
Increase in Var(ε) due to:
Interest rate
90D CDOR
BAX1
BAX2
Var(ε) on
non-comm days
4.608
(0.000)
17.291
(0.000)
29.561
(0.000)
Fad Press
Release
1.137
(0.442)
18.345
0.004
31.538
(0.013)
19
MPR(U)
Speeches
11.486
(0.278)
30.053
(0.297)
37.011
(0.270)
-0.010
(0.995)
11.264
(0.079)
27.197
(0.036)
Although the MPR has a significant impact on 10 year Bond, the coefficient is negative, which seemed a
bit strange at first, however, the BoC attemps to set MPR dates when there are not other major economic
events, therefore it is possible that if the MPR has no impact on the volatility of 10 year yields, on average
MPR dates may have a lower variance than other non-communication dates.
25
BAX3
2Y Bond
5Y Bond
10Y Bond
35.703
(0.000)
23.929
(0.000)
22.129
(0.000)
18.452
(0.000)
34.532
(0.304)
23.584
(0.280)
12.222
(0.336)
3.955
(0.651)
34.880
(0.026)
19.596
(0.058)
5.942
(0.448)
-1.299
(0.775)
27.055
(0.071)
4.779
(0.491)
0.392
(0.933)
-0.799
(0.827)
Table 2b: Second Subsample Results
Increase in Var(ε) due to:
Interest rate
90D CDOR
BAX1
BAX2
BAX3
2Y Bond
5Y Bond
10Y Bond
Var(ε) on
non-comm days
0.923
(0.000)
8.012
(0.000)
15.889
(0.000)
19.110
(0.000)
9.906
(0.000)
10.054
(0.000)
9.376
(0.000)
Fad Press
Release
0.308
(0.441)
8.423
0.070
13.509
(0.121)
13.798
(0.175)
7.956
(0.161)
3.007
(0.415)
0.113
(0.969)
MPR(U)
Speeches
-0.124
(0.733)
1.348
(0.760)
9.329
(0.485)
12.077
(0.462)
4.607
(0.528)
-2.803
(0.193)
-4.942
(0.001)
1.100
(0.199)
1.417
(0.581)
1.597
(0.680)
1.389
(0.753)
3.020
(0.409)
2.814
(0.450)
0.952
(0.727)
These findings seem to support the idea that markets focus nearly solely on the FLS and
view it as a rough pre-commitment in that, in contrast to our earlier results, FAD press
statements are no longer significant. However, it could also be the case that markets
reduced reaction to FAD press releases is due to their better or increased understanding of
the reaction function of the Bank, as markets became accustomed to the new FAD
regime. That is, there are fewer information asymmetries between the central bank and
markets about the reaction function and therefore less ‘new’ information in central bank
communication. We will look at each of these explanations in turn.
26
One caveat before we continue is that there are two factors that may reduce the
robustness of these results somewhat. The first is that because we are splitting our
sample, our sample size is halved, reducing the statistical strength of the test. The second
is that there are significantly fewer periods of uncertainty in the second half (and fewer
macroeconomic turning points in monetary policy) and as such market participants
perhaps have less “new” or less “important” information to react to, relative to the earlier
sample.
7.1 Testing Possible Explanations
To test whether central has become more transparent or just more predictable over the
second half of our sample we conduct two tests. In the first, we conduct a test using a
cross dummies which are first set to one on FAD press release dates that contain a FLS,
and, separately, the cross dummies are set to one on the subsequent FAD date (the FAD
date following one that has a FAD press release with a FLS). An additional explanatory
variable, the variable that cross-multiplies the original FAD dummy with one of the two
FLS dummy, is added to the second stage regression. If markets are ignoring information
surrounding the FLS we would expect the coefficient on our cross-dummy variables to be
large and negative.
Table 3 provides the coefficients and their significance level for the full sample and cross
dummy for both the current FAD press statement and the subsequent FAD press
statement scenarios. As can be seen below, the coefficients on the cross dummies are in
general negative and, at least in the first scenario significant at the 10 per cent level for all
but one of the yields. This supports our hypothesis that the Bank of Canada has become
more predictable over the second half of our sample.
Table 3: FAD FLS Cross Dummy Regressions
Interest rate
90D CDOR
Testing the impact of the FLS on FAD Testing the impact of FLS on
Press Statement Days
Subsequent FAD Press Statement
Days
FAD Press Release Fad Press Release FAD Press
Subsequent FAD
Full Sample
Cross Dummy
Release Full
Press Release
Sample
Cross Dummy
2.741
(0.084)
2.482
(0.102)
-3.865
(0.021)
27
-3.491
(0.035)
BAX1
BAX2
BAX3
2Y Bond
5Y Bond
10Y Bond
22.607
(0.001)
42.049
(0.003)
50.078
(0.003)
26.968
(0.028)
13.705
(0.101)
3.888
(0.437)
-15.406
0.065
-33.610
(0.040)
-46.150
(0.018)
-23.083
(0.073)
-16.474
(0.079)
-7.967
(0.180)
14.155
(0.022)
26.461
(0.034)
35.355
(0.070)
19.728
(0.058)
9.062
(0.279)
1.385
(0.776)
1.237
(0.886)
-3.127
(0.854)
-17.934
(0.375)
-9.513
(0.463)
-7.837
(0.405)
-3.274
(0.582)
In the second test we create a 2nd half Canadian and US macro surprise cross dummy by
creating a set of dummy variables for macro releases in the second half only, and rerun
regression 1 with the inclusion of this new dummy. If markets understand the central
bank’s reaction function better, Canadian macro surprise cross dummy test should yield
significant positive coefficients as market participants react more fully to new domestic
economic information as it arrives. We find however, that for all yields, the majority of
the macro surprise dummies were negative20, suggesting that markets reacted less to
Canadian macro releases in the second half of our sample. Thus lending further support
to our increased predictability hypothesis.
8. Conclusion
We find evidence that official Bank of Canada communication, in particular FAD press
statements have a significant effect on near to medium term yields suggesting that these
communications convey important new information which impact the markets interest
rate expectations. The lack of significance in longer term yields may suggest a high level
of credibility surrounding the Bank of Canada’s single inflation target objective of 2 per
cent over our period of study. As well, the observation that MPR(U)’s were not
significant did not come as surprise for two reasons. Over this period, the MPR(U)’s
closely followed FAD press statements and by construction conveyed a consistent
message for the near to medium term economic outlook, and as mentioned above, longterm yields appear to be well anchored over this period in Canada.
20
A number of these negative cross dummies were also significant at the 5 per cent level. As well, none of
the cross dummies with positive coefficients were significant at the 5 per cent level.
28
Over our period of study, in an effort to increase the level of transparency, there was a
major shift in the Bank of Canada’s communication strategy, the inclusion of forward
looking statements in FAD press statements and MPR(U)’s, which allows us to study the
impact of these statements on Central Bank transparency. Reruning our tests on both a
split sample (where these statements began being used in a consistent manner) as well as
cross-dummies macro-variables to account for dates prior to our split that include FLS’s,
our results suggest that there has been a loss of conditionality surrounding forward
looking monetary policy assumptions with the introduction of these forward looking
statements.
Care must be taken, however, when interpreting results of empirical work studying the
effects of forward looking language. There are a number of issues related to the small
sample size as well as lack of different economic environments. For instance, there are
only a few policy turning points over our full sample and none in the second half of
sample, the period when forward looking statements were consistently used. As such,
there is less uncertainty as well as fewer macroeconomic shocks and/or news to react to,
possibly contributing to some of our second half results. As well, empirical work
suggests that the pre-existing shape of the yield curve at the time of the communication
will impact how markets react to news along the yield curve.
Another caveat is that the Bank of Canada has stressed in its communications that it does
not react to any one macroeconomic shock or surprise. The smaller reaction of market
rates in the second half of our sample to macroeconomic news may in part be a reflection
of the market’s better understanding how the Bank of Canada reacts to these shocks.
Consequently, instead of reacting to one off shocks, there is more of a gradual shift in
policy rate expectations with the accumulation of data that we are unable to pick up in
our tests.
Finally, using data at a daily frequency may also affect our results as it is not possible to
control for all other shocks hitting the market the same day, and further study at an
intraday trading frequency might yield different answers.
29
That said, there is a general agreement amongst central bankers that in general, there
remain issues surrounding the incorporation of conditionality and uncertainty around this
form of guidance, the debate surrounds the weighting of the risks versus the benefits, and
the various views on how conditionality can be incorporated into the communications
strategy. As such, each central bank has its own view of both the risks and benefits, as
well as how to overcome the lack of conditionality surrounding these statements
(projections). Consequently, there exists a full spectrum of communication strategies
surrounding how much of the policy outlook to reveal, from not saying anything (except
in extreme circumstances), to publishing regularly a policy rate forecast. There may also
be no one “ideal” communications strategy for mitigating risks to conditionality and
uncertainty surrounding these statements.
30
Bibliography
Andreou, J. 2005. “Estimating the Impact of Monetary Policy Surprises on Fixed-Income
Markets.” Bank of Canada Review (Summer): 11-19.
Connolly, E and M. Kohler. 2007. “News and Interest Rate Expectations: A Study of Six
Central Banks”. In The analysis of the money market: role, challenges and implications
from the monetary policy perspective. Proceedings of a conference held by the European
Central Bank. November 2007. Frankfurt: European Central Bank.
Ehrmann, M. and M. Fratzscher. 2005. “How Should Central Banks Communicate?”
European Central Bank Working Paper No. 557.
Ehrmann, M. and M Fratzscher. 2007a. “Communication and Decision-Making by
Central Bank Committees: Different Strategies, Same Effectiveness?” Journal of Money,
Credit and Banking, 39(2–3): 509-41.
Ehrmann, M. and M. Fratzscher. 2007b. “Transparency, disclosure and the Federal Reserve.”
International Journal of Central Banking 3(1): 179-225.
Gravelle, T. and R. Moessner. 2002. “Reactions of Canadian Interest Rates to
Macroeconomic Announcements: Implications for Monetary Policy Transparency.”
Journal of Bond Trading and Management (1) 1: 27-43.
Harvey, N. 1996. “The Market for Futures Contracts on Canadian Bankers’
Acceptances.” Bank of Canada Review (Autumn): 19-36.
Jen, S. “Currency Economics: My Thoughts on Currencies.” Morgan Stanley, 26
February, 2007.
Johnson, G. 2003. “Measuring Interest Rate Expectations in Canada.” Bank of Canada
Review (Summer): 17-27.
Johnson, M., G. Johnson, and C. Fay. 2006. “Communicating with Financial Markets.”
MPRC Discussion Note.
Johnston, J., and J DiNardo.1997. Econometric Methods, Fourth Edition. The McGrawHill Companies, Inc.
Kahn, G. 2007. Communicating a Policy Path: The Next Frontier in Central Bank
Transparency? Kansas City Economic Review, First quarter 2007.
King, M. 2006. Speech at the Lord Mayor’s Banquet for Bankers and Merchants of the
City of London at the Mansion House, London, England, 21 June.
Kohn, D. and B. Sack. 2003. “Central Bank Talk: Does it Matter and Why?” Board of
Governors of the Federal Reserve System.
31
Kohn, D. 2005. “Central Bank Communication.” Speech at the Annual Meeting of the
American Economic Association, Philadelphia, Pennsylvania. January 9. Federal Reserve
Board of Governors.
Kuttner, K. 2000. “Monetary Policy Surprises and Interest Rates: Evidence from the Fed
Funds Futures Market.” Federal Reserve Bank of New York Staff Report No.99.
Moessner R., T. Gravelle, and P. Sinclair. 2005. “Measures of Monetary Policy
Transparency and the Transmission Mechanism”. In Mahadeva, L. & P. Sinclair (eds.).
How Monetary Policy Works. Routledge.
Moessner, R. and W. Nelson. 2008. "Central Bank Policy Rate Guidance and Financial
Market Functioning" . BIS Working Paper No. 246
Muller, P. and M. Zelmer. 1999. “Greater Transparency in Monetary Policy: Impact on
Financial Markets.” Bank of Canada Technical Report No. 86.
Parent, N. 2002-2003. “Transparency and the Response of Interest Rates to the
Publication of Macroeconomic Data” Bank of Canada Review (Winter): 29-34.
Reeves, R. and M. Sawicki. 2007. “Do Financial Markets React to Bank of England
Communication?” European Journal of Political Economy 23: 207-227.
Reinhart, V. and B. Sack. 2006. “Grading the Federal Open Market Committee’s
Communications.” Paper presented at the 2006 AEA meetings, Boston.
Rudebusch, Glenn D. 2008. “Publishing Central Bank Interest Rate Forecasts.” FRBSF
Economic Letter No. 2008-02.
32
Appendix A: Control Variables
A.1 Macroeconomic Surprises
Gravelle & Moessner (2002) and Parent (2002-2003), among others, found that the
surprise component of macroeconomic announcement has a statistically significant
impact on a number of Canadian market rates. Following Gravelle & Moessner (2002)
and Reeves & Sawicki (2005), we calculate standardized21 macroeconomic surprises as:
mac =
u
i ,t
( X i ,t − X ie,t )
Ω iX
(A1)
Where X i ,t − X ie,t is the actual minus the market expected value of the ith macroeconomic
release on day t, and Ω iX is the sample standard deviation of surprises for the ith
macroeconomic release. This is set to zero on days where no announcements are made.22
Financial market expectations or forecast of the macroeconomic data release used in
calculating the surprise component are provided by Bloomberg surveys conducted prior
to each announcement. We use Bloomberg market poll surveys as our market
expectations instead of Money Market Services International as this latter source has data
limitations in the more recent period; however, both polls survey a similar group of
economists2324.
21
We standardize the macroeconomic surprises in order to make the coefficients comparable across
indicators.
22
Another factor that would influence yields would be revisions to macroeconomic releases. However, we
do not have a full sample of data for this, and so will not include it in our study. Given that the macro
revision is released at the same time as the current period macro release, not including the revisions if
anything may make the result marginally noisier.
23
Although Gravelle & Moessner (2002) and Parent (2002-2003) use S&P MMS survey data for their full
sample, we were unable to do this as the survey becomes inconsistent after mid. 2003 Recent empirical
work has faced this same challenge and has found that where the two surveys overlap, the two data sources
agree very closely. Comparison of our data confirm this.
24
Reeves & Sawicki also split their macroeconomic survey data sample between MMS & Bloomberg due
to data limitations. Other papers that have dealt with this issue include Robitaille & Roush (2005),
Gurkaynak, Levin, & Swanson (2006), and Gurkaynak & Wolfers (2005).
33
A.2 Policy Surprises
As pointed out in Kohn & Sack (2003) and empirically shown through factor analysis in
Gurkaynak, Sack and Swanson (2004), the market response to a policy announcement
consists of two distinct factors: The first is the difference between where markets expect
the central bank to set its policy rate and the action the Central Bank actually takes on
one particular policy date. This is the “surprise” component. The second is linked to the
press statement accompanying the policy action and is the perceived change in the future
path of policy.
As we want to focus on the communications aspect, the press statement, we will need to
control for any impact the “surprise” component of the policy action by the Bank of
Canada has on our key rates. Given our close link with the US, we will also test for
Canadian market reaction to US policy surprises.
A.2.1 Canadian Policy Surprises
Following the methodology of Andreou (2005) we will take the one day difference in the
1 month bankers acceptance (BA) rate25 on Canadian monetary policy decision days.
ΔON tu = bat − bat −1
(A2)
Where bat- bat-1 is the surprise component of policy. We set this equal to zero on nonpolicy days.
A.2.2 US Policy Surprises
Consistent with the work of Kuttner (2000) we measure US policy surprises as:
Δff t u = [D /( D − d )] ⋅ Δff 1t
25
(A3)
Johnson (2003) shows empirically, by testing the expectations hypothesis, that 1 month Bankers
Acceptances are the best proxy for market expectations (over our period of study). A number of studies use
the Reuters survey of market economists as their proxy for market expectations. We find this measure in
Canada to not be statistically significantly different from our measure, however we chose our measure for a
number of reasons including the time lag between the survey and the actual announcement.
34
Where D is the total # of days in the month, d is the day of the month of the FOMC
decision, and Δff 1t is the change in the futures rate on the day of the policy decision
(including inter-meeting actions). We set this equal to zero on non-policy days.
A.3 U.S. Communications
As pointed out in Gravelle & Moessner (2002) and shown empirically in Connolly &
Kohler (2004), “Because Canada is a small open economy, with direct links to the US
economy in terms of trade and capital flows, it should be of no surprise to find that
Canadian Debt instruments are significantly influenced by US interest rates, which are in
turn affected by various US economic announcements”. Although we have already
included US macro announcements and policy surprises, we have not controlled for
another major factor that influences US interest rates (and in turn Canadian rates), FOMC
communications.
To control for the impact of FOMC communications on Canadian rates we will include
the one day change in the second Eurodollar futures contract26 as well as the one day
change in the on the run 2 year Treasury on dates of FOMC press releases, testimonies
and minutes27.
26
The second Eurodollar futures contract is tied to the three-month LIBOR rate on the date of expiration –
a rate that is primarily influenced by the expected fed funds rate over the subsequent three months.
27
Reinhart & Sack (2005) test both short and medium term expectations using these rates and find FOMC
press releases, testimonies and minutes (following the change in release time) to be significant. We will
therefore only include these Fed communications into our regression.
35
Appendix B: The Role of Fixed Announcement Dates in Enhancing Monetary Policy
Reaction Function Understanding
As argued in Gravelle and Moessner (2002) and Moessner, Gravelle, and Sinclair (2005)
a key ingredient to greater monetary policy reaction function understanding by market
participants is fixed pre-announced policy rate action dates. Without, fixed action dates
market participants find it difficult accumulate and distil the macroeconomic data in a
fashion that roughly simulates the way the Bank uses it for its economic outlook and as
such properly understand a central banks monetary policy reaction function.
The
transparency advantages of fixed action dates are brought out in stark relief when one
thinks about fixed dates on which a central bank does not change the policy rate. In a
regime without fixed dates, a day that bank does not change rates looks the same as any
other day. However, when fixed dates are in place the fact the central bank did not
change its policy rate on the designated date is news.
In addition to information
generated by the action itself (of moving the target rate or not on fixed action dates),
fixed action dates also provide the central bank with an additional opportunity to
communicate to the public, via a press release that accompanies the monetary policy
decision point, its views about its economic outlook and what drives any revision to it as
well as its policy stance or guidance that results from this outlook.
36
Appendix C
Table C.1: Use of Forward Looking Policy Language in Bank of Canada FAD Press
Statements and MPR(U)’s
Date
Event
Change
in the
Target
ON Rate
Near Term
Direction of
ON Rate
Implied by FLS
Balance
of Risk
Forward-Looking Policy
Statement (FLS)
Balance of Risks
Statement
May
29,
2007
FAD
0
Higher ON
rate
Slight tilt
to upside
… some increase in the
target for the overnight rate
may be required in the near
term…
On balance, the Bank
judges that there is an
increased risk that
future inflation will
persist above the 2 per
cent inflation target…
Apr 26,
2007
MPR
Unchanged
ON rate
Slight tilt
to upside
The current level of the
target for the overnight rate
is judged, at this time, to be
consistent with achieving
the inflation target over the
medium term.
The Bank continues to
judge that the risks to
its inflation projection
are roughly balanced,
although there is now a
slight tilt to the upside.
Apr 24,
2007
FAD
0
Unchanged
ON rate
Slight tilt
to upside
Mar 6,
2007
FAD
0
Unchanged
ON rate
Balanced
The Bank continues to
judge that the risks to
its inflation projection
are roughly balanced,
although there is now a
slight tilt to the upside.
Despite recent volatility
in global financial
markets, the Bank
continues to judge that
the risks to its inflation
projection are roughly
balanced
Jan 18,
2007
MPRU
Unchanged
ON rate
Balanced
The current level of the
target for the overnight rate
is judged, at this time, to be
consistent with achieving
the inflation target over the
medium term.
In line with the Bank's
outlook, the current level of
the target for the overnight
rate is judged, at this time,
to be consistent with
achieving the inflation
target over the medium
term.
The current level of the
policy interest rate is
judged, at this time, to be
consistent with achieving
the inflation target.
Jan 16,
2007
FAD
0
Unchanged
ON rate
Balanced
In line with the Bank’s
outlook, the current level of
the target for the overnight
rate is judged, at this time,
to be consistent with
achieving the inflation
target over the medium
term.
Dec 5,
FAD
0
Unchanged
Balanced
In line with the Bank's
37
The risks around the
Bank's inflation
projection continue to
be judged to be roughly
balanced, but the main
upside and downside
risks have diminished
somewhat since the
October MPR.
The Bank continues to
judge that the risks to
the inflation projection
are roughly balanced,
but the main upside and
downside risks outlined
in the October MPR
have diminished
somewhat
The Bank judges that,
2006
ON rate
Oct 19,
2006
MPR
Oct 17,
2006
FAD
Sep 6,
2006
FAD
Jul 13,
2006
MPRU
July
11,
2006
FAD
May
24,
2006
FAD
Apr 27,
2006
MPR
Unchanged
ON rate
Balanced
0
Unchanged
ON rate
Balanced
0
Unchanged
ON rate
Balanced
Unchanged
ON rate
Small tilt
to
downside
later in
projection
0
Unchanged
ON rate
Small tilt
to
downside
later in
projection
+25bp
Unchanged
ON rate
Short term
risks
balanced.
Small tilt
to
downside
later in
projection
Higher ON
Rate
Short term
risks
balanced.
Small tilt
38
outlook, the current level of
the target for the overnight
rate is judged at this time to
be consistent with achieving
the inflation target over the
medium term.
The current level of the
policy interest rate is
judged, at this time, to be
consistent with achieving
the inflation target.
In line with this updated
outlook, the current level of
the target for the overnight
rate is judged at this time to
be consistent with achieving
the inflation target over the
medium term.
In line with this outlook, the
current level of the target
for the overnight rate is
judged at this time to be
consistent with achieving
the inflation target over the
medium term.
The current level of the
policy interest rate is judged
at this time to be consistent
with achieving the inflation
target
In line with the Bank's
largely unchanged outlook,
the current level of the
target for the overnight rate
is judged at this time to be
consistent with achieving
the inflation target over the
medium term.
With today's increase, the
target for the overnight rate
is now at a level that is
expected to keep the
Canadian economy on the
base-case path projected in
the April Monetary Policy
Report (MPR) and to return
inflation to the 2 per cent
target.
In line with the Bank's
outlook, some modest
further increase in the policy
interest rate may be
overall, risks around the
inflation projection are
roughly balanced.
The Bank judges that
the risks to its inflation
projection are roughly
balanced.
It is the Bank's
judgment that, overall,
risks around the
inflation projection are
roughly balanced.
While both these risks
appear to be a little
greater than they were
in July, the Bank
continues to judge that,
overall, risks are
roughly balanced.
the Bank continues to
judge that these risks
are roughly balanced,
with a small tilt to the
downside later in the
projection period
because of the
possibility of a
disorderly resolution of
global imbalances.
Risks to the projection
remain roughly
balanced, with a small
tilt to the downside
later in the projection
period related to global
imbalances.
The Bank continues to
assess the risks to this
projection to be as
presented in the MPR.
The Bank judges that
these risks are roughly
balanced, with a small
tilt to the downside
to
downside
later in
projection
Short term
risks
balanced.
Small tilt
to
downside
later in
projection
Apr 25,
2006
FAD
+25bp
Higher ON
Rate
Mar 7,
2006
FAD
+25bp
Higher ON
Rate
Jan 26,
2006
MPRU
Jan 24,
2006
FAD
+25bp
Higher ON
rate
Short term
risks
balanced.
Risks
tilted to
downside
later in
projection
Dec 6,
2005
FAD
+25bp
Higher ON
rate
Oct 20,
2005
MPR
Short term
risks
balanced.
Risks
tilted to
downside
later in
projection
Short term
risks
balanced.
Risks
tilted to
Higher ON
Rate
Higher ON
rate
Short term
risks
balanced.
Risks
tilted to
downside
later in
projection
Short term
risks
balanced.
Risks
tilted to
downside
later in
projection
39
required.
later in the projection
period.
In line with the Bank's
outlook for the Canadian
economy, some modest
further increase in the policy
interest rate may be required
to keep aggregate supply
and demand in balance and
inflation on target over the
medium term.
Consistent with this view,
some modest further
increase in the policy
interest rate may be required
to keep aggregate supply
and demand in balance and
inflation on target over the
medium term
In line with the Bank's
outlook, some modest
further increase in the policy
interest rate would be
required.
The Bank judges that
the risks to its
projection are roughly
balanced, with a small
tilt to the downside
later in the projection
period.
In line with the Bank's basecase projection and current
assessment of risks, some
modest further increase in
the policy interest rate
would be required to keep
aggregate supply and
demand in balance and
inflation on target over the
medium term.
In line with the outlook,
some further reduction in
monetary stimulus will be
required to maintain a
balance between aggregate
supply and demand over the
next four to six quarters and
keep inflation on target.
In line with the Bank's
outlook, some further
reduction of monetary
stimulus will be required.
Risks to the Bank's
projection remain
balanced for 2006 and
tilted to the downside
through 2007 and
beyond.
Recent data do not alter
the Bank's outlook for
growth and inflation,
including its assessment
of risks, as set out in the
January Update
Risks to the Bank's
projection remain
balanced for 2006.
Through 2007 and
beyond, risks are tilted
to the downside, as the
unwinding of global
imbalances could
involve a slowdown in
world economic
activity.
The Bank continues to
judge that the risks to
the outlook are
balanced over the short
term, but are tilted to
the downside through
2007 and beyond.
Short-term risks to this
projection appear to be
balanced. But looking
further out to 2007 and
beyond, there are
downside
later in
projection
Oct 18,
2005
FAD
+25bp
Sep 7,
2005
FAD
+25bp
Jul 14,
2005
MPRU
Jul 12,
2005
FAD
May
25,
2005
FAD
Higher ON
rate
Short term
risks
balanced.
Risks
tilted to
downside
later in
projection
Short term
risks
balanced.
Risks
tilted to
downside
later in
projection
In line with the Bank's
outlook, and given that the
Canadian economy now
appears to be operating at
capacity, some further
reduction of monetary
stimulus will be required to
maintain a balance between
aggregate supply and
demand over the next four
to six quarters, and to keep
inflation on target.
None
increasing risks that the
unwinding of global
economic imbalances
could involve a period
of weak world
economic growth.
Short-term risks to this
projection appear to be
balanced. But as we
look further out to 2007
and beyond, there are
increasing risks that the
unwinding of global
economic imbalances
could involve a period
of weak global growth.
Despite developments
associated with higher
energy prices, risks to
the Bank's outlook for
the Canadian economy
through 2006 still
appear to be reasonably
balanced. Over the
medium term, however,
there is increasing risk
that the correction of
global current account
imbalances could
involve a period of
weakness in world
aggregate demand.
These risks appear to be
balanced. Over the
medium term, however,
there is increasing risk
that the correction of
global current account
imbalances could
involve a period of
weakness in world
aggregate demand.
Higher ON
rate
Tilted to
downside
later in
projection
In line with this outlook,
some reduction of monetary
stimulus will be required in
the near term.
0
Higher ON
rate
Tilted to
downside
later in
projection
However, in line with the
Bank's outlook, some
reduction in the amount of
monetary stimulus will be
required in the near term to
keep aggregate demand and
supply in balance and
inflation on target.
The risks to the outlook
through 2006 appear
balanced, but over the
medium term risks
related to global
imbalances are
increasing.
0
Higher ON
rate
In line with this outlook, a
reduction of monetary
stimulus will be required
None
40
over time.
Apr 14,
2005
MPR
Apr 12,
2005
FAD
Mar 1,
2005
FAD
Jan 27,
2005
MPRU
Higher ON
rate
In line with this outlook, a
reduction of monetary
stimulus will be required
over time.
None
0
Higher ON
rate
None
0
Higher ON
rate
In line with this outlook, a
reduction of monetary
stimulus will be required
over time.
…the implications for the
pace of reduction in
monetary stimulus are
essentially unchanged from
those that the Bank
presented in January's
Update
The pace of reduction in
monetary stimulus is likely
to be slower than envisioned
in the October Report.
Higher ON
rate (but at a
slower rate than
anticipated at the
time of the MPR)
None
None
Jan 25,
2005
Dec 7,
2004
Oct 21,
2004
FAD
0
None – refers to MPRU
None
FAD
0
None
None
None
Oct 19,
2004
FAD
+25bp
Higher ON
rate
Sep 8,
2004
Jul 22,
2004
FAD
+25bp
Higher ON
rate
Higher ON
rate
This base-case projection
assumes further reduction of
monetary stimulus over time
to keep the economy near its
production capacity and
achieve the inflation target.
Further reduction of
monetary stimulus will be
required over time to keep
inflation on target, with the
pace depending on the
Bank's continuing
assessment of the prospects
for factors that affect
pressures on capacity and,
hence, inflation.
None
None
Jul 20,
2004
Jun 8,
2004
Apr 15,
FAD
0
Monetary stimulus will have
to be removed to avoid a
buildup of inflation
pressures. The pace of the
withdrawal will depend on
the evolving prospects for
inflation and for capacity
pressures
None – refers to MPRU
FAD
0
None
None
None
Overall, the risks to the
MPR
Higher ON
rate
MPRU
MPR
Balanced
41
None
None
None
2004
Apr 13,
2004
Mar 2,
2004
Jan 22,
2004
Jan 20,
2004
Dec 2,
2003
Oct 22,
2003
Oct 15,
2003
Sep 3,
2003
Jul 17,
2003
July
15,
2003
June 3,
2003
Apr 23,
2003
FAD
-25bp
Balanced
FAD
-25bp
Apr 15,
2003
Mar 4,
2003
FAD
+25bp
FAD
+25bp
Jan 23,
2003
MPRU
Jan 21,
2003
FAD
0
Higher ON
rate
Dec 3,
FAD
0
Higher ON
MPRU
None
outlook appear
balanced.
The risks to the outlook
now appear balanced
None
None
None
None
FAD
-25bp
None
None
FAD
0
None
None
None
None
MPR
FAD
0
None
None
FAD
-25bp
None
None
None
None
MPRU
FAD
-25bp
None
None
FAD
0
None
None
The Bank believes that
further reductions in
monetary stimulus over time
will be necessary, but the
timing and pace will depend
on the evolution of inflation
expectations and the
strength of domestic and
external demand.
None
None
As indicated in the MPR
Update, further reductions
in monetary stimulus will be
required to return inflation
to the target over the
medium term.
With the stance of monetary
policy currently very
stimulative, a reduction of
stimulus will be required in
order to return inflation to
the 2 per cent target over the
medium term.
However, with the stance of
monetary policy currently
very stimulative, a reduction
of stimulus will be required
in order to return inflation to
the 2 per cent target over the
medium term.
timely removal of monetary
None
MPR
Higher ON
rate
Higher ON
rate
Higher ON
rate
42
None
None
None
None
2002
rate
Oct 23,
2002
MPR
Higher ON
rate
Oct 16,
2002
FAD
0
Higher ON
rate
Sep 4,
2002
FAD
0
Higher ON
rate
Jul 24,
2002
MPRU
Jul 16,
2002
Jun 4,
2002
Apr 24,
2002
FAD
+25bp
FAD
+25bp
Apr 16,
2002
Mar 5,
2002
Jan 23,
2002
Jan 15,
FAD
FAD
Higher ON
rate
Balanced
None
None
At this time, the risks to
our projected rate of
growth of 3 to 4 per
cent growth appear to
be balanced. (PR)
None
None
None
+25bp
As the economy approaches
its capacity, the task for
monetary policy is to gauge
economic strength and the
implications for future
inflation. This means
reducing the substantial
amount of stimulus in place
in a timely and measured
manner.
None
0
None
None
None
None
None
None
MPRU
FAD
None
None
MPR
Higher ON
rate
Higher ON
rate
stimulus will be required to
achieve the inflation target
over the medium term
Going forward, further
removal of monetary
stimulus will be required,
with the pace and extent of
the tightening depending on
unfolding developments and
on their implications for
pressures on capacity and
inflation in Canada.
It remains the Bank's view,
going forward, that timely
removal of monetary
stimulus will be required to
achieve the inflation target
over the medium term.
Looking forward, it remains
the Bank's view that, as the
Canadian economy
continues to expand and to
approach its production
capacity, further measured
reductions in monetary
stimulus will be necessary
in order to achieve the
inflation control target of 2
per cent over the medium
term.
It remains the Bank's view
that the underlying
economic situation will
require further reductions in
the amount of monetary
stimulus (PR)
None
-25bp
43
None
2002
Nov
27,
2001
Nov 7,
2001
Oct 23,
2001
Sep 17,
2001
Aug
28,
2001
Aug 1,
2001
Jul 27,
2001
May
29,
2001
May 1,
2001
April
17,
2001
Mar 6,
2001
Feb 6,
2001
Jan 23,
2001
Dec 5,
2000
Nov 9,
2000
FAD
-50bp
MPR
None
None
None
None
FAD
-75bp
None
None
Intermeeting
FAD
-50bp
None
None
-25bp
None
None
None
None
MPRU
FAD
-25bp
None
None
FAD
-25bp
None
None
None
None
MPR
FAD
-25bp
None
None
FAD
-50bp
None
None
None
None
MPRU
FAD
-25bp
None
None
FAD
0
None
None
None
None
MPR
Table C.2a: Consistency of BOC forward looking policy statements and monetary
policy decisions: Post-FAD full sample excld. 3 months following September 11
Monetary Policy Decision (following meeting)
Direction of Forward
looking Policy Statement
Easing
No Change
Tightening
All
in Press Release or
MPR(U)
0
0
0
0
Easing
0
8
0
8
Neutral
0
10*
12
22
Tightening
12
6
2
20
No Forward Looking
Statement
12
25
14
50
All
44
*on one occasion there was no change in the following meeting but an ease in the next
meeting
Table C.2b: Consistency of BOC forward looking policy statements and monetary
policy decisions: Post-FAD 2nd half
Monetary Policy Decision (following meeting)
Direction of Forward
looking Policy Statement
Easing
No Change
Tightening
All
in Press Release or
MPR(U)
0
0
0
0
Easing
0
8
0
8
Neutral
0
5
10
15
Tightening
0
0
0
0
No Forward Looking
Statement
0
13
10
23
All
45
Table C.3: Canadian interest rate response to surprises: pre and post FAD
subsamples
Rate
Front
BAX
Significant surprises,
Post-FAD (Oct.31/00May 31/07);
c_cpixfe (2.2660, 0.0000);
c_emp (2.2210, 0.0006);
c_gdp (0.8430, 0.0110);
c_ivey (0.8356, 0.0224);
c_rtl_sls (1.6918, 0.0022)
2
R
0.3024
0.2694
us_cpixfe (1.0142, 0.0292);
us_gdp (1.8894, 0.0054);
us_mich (0.8012, 0.0389);
us_nfp (5.1355, 0.0000);
us_unemp (-1.7781, 0.0073)
us_cpixfe (1.4655, 0.0074);
us_gdp (2.2557, 0.0039);
us_hr_earn (2.1960, 0.0179);
us_ism (2.3627, 0.0376);
us_nfp (6.4945, 0.0000);
us_unemp (-1.9770, 0.0111)
0.2568
c_cpixfe (2.5352, 0.0009);
c_emp (3.2099, 0.0004);
c_gdp (1.6745, 0.0048);
c_rtl_sls (2.3064, 0.0041)
2-year
Bond
5-year
Bond
us_cpixfe (1.1361, 0.0171);
us_gdp (1.6855, 0.0151);
us_hr_earn (1.9372, 0.0079);
us_ind_prod (0.9930, 0.193);
us_ism (2.2035, 0.0131);
us_nfp (5.4516, 0.0000);
us_unemp (-1.8403, 0.0066)
c_cpixfe (1.8511, 0.0055);
c_emp (2.4162, 0.0027);
c_hsg_sts (-0.9512, 0.0149);
c_ivey (1.2873, 0.0474);
c_rtl_sls (1.6627, 0.0195)
0.3585
0.2634
0.2056
us_cpixfe (1.6574, 0.0260);
us_gdp (2.5225, 0.0009);
us_hr_earn (2.0707, 0.0469);
us_ind_prod (1.9635,
0.0218);
us_ism (2.6915, 0.0440);
us_mich (1.3477, 0.0163);
us_nfp (5.6011, 0.0000)
c_cpixfe (3.0472, 0.0194);
c_emp (4.6898, 0.0062);
c_rtl_sls (3.8159, 0.0083)
us_cpixfe (2.4493, 0.0034);
us_gdp (2.7636, 0.0059);
us_hr_earn (2.8730, 0.0178);
us_ind_prod (2.1186,
0.0276);
us_ism (3.1486, 0.0277);
us_mich (1.5479, 0.0252);
us_nfp (6.9083, 0.0000)
c_cpixfe (2.4617, 0.0257);
c_emp (3.1078, 0.0311);
c_ivey (2.0236, 0.0400);
c_rtl_sls (2.7872, 0.0332)
us_cpixfe (1.6800, 0.0147);
us_gdp (2.0822, 0.0190);
us_hr_earn (2.5396, 0.0147);
us_ind_prod (2.2551,
0.0004);
us_ism (3.6300, 0.0079);
us_nfp (5.6645, 0.0000);
us_unemp (-1.6721, 0.0421)
c_cpixfe (1.9462, 0.0419);
c_curr_acct (-1.9733,
0.0144);
c_hsg_sts (-1.3573, 0.0185);
c_ivey (2.4559, 0.0084)
Significant surprises,
Post-FAD 2nd Half (July
22/04-May31/07)
c_cpixfe (1.4575, 0.0080);
c_gdp (1.4019, 0.0002);
c_ivey (0.7757, 0.0373);
c_raw_mat (-0.4882, 0.0429)
0.3091
c_cpixfe (2.6606, 0.0063);
c_emp (2.7756, 0.0049);
c_gdp (2.3579, 0.0005);
c_rtl_sls (1.6396, 0.0405)
0.2899
c_cpixfe (2.6648, 0.0103);
c_emp (3.9098, 0.0001);
c_gdp (2.9563, 0.0006);
c_rtl_sls (1.7852, 0.0196)
0.2394
0.2500
us_nfp (7.6713, 0.0000);
us_ppixfe (2.0647, 0.0000);
us_unemp (-4.8649, 0.0093)
0.2909
c_cpixfe (2.5743, 0.0036);
c_emp (3.3618, 0.0000);
c_gdp (2.2952, 0.0024);
c_rtl_sls (1.6200, 0.0106)
0.2706
us_nfp (6.1859, 0.0000);
us_ppixfe (0.8722, 0.0437);
us_unemp (-3.2702, 0.0167)
0.2234
c_curr_acct (3.0674,
0.0187);
c_emp (3.0067, 0.0000);
c_rtl_sls (1.2659, 0.0380)
The first number in parentheses is the coefficient; the second number represents the significance level.
Estimated using a Newey-West adjusted covariance matrix.
46
0.2199
us_nfp (6.1014, 0.0000);
us_ppixfe (1.5567, 0.0002);
us_unemp (-4.0079, 0.0075)
us_nfp (5.3438, 0.0000);
28
R2
us_nfp (3.1476, 0.0000);
us_ppixfe (0.4860, 0.0421);
us_unemp (-2.2254, 0.0050)
c_cpixfe (2.4025, 0.0123);
c_cpi (1.8341, 0.0318);
c_emp (4.8005, 0.0010);
c_rtl_sls (3.5724, 0.0079);
c_cpixfe (2.5208, 0.0008);
c_emp (3.9619, 0.0000);
c_gdp (1.7439, 0.0080);
c_ivey (1.2418, 0.0384);
c_rtl_sls (2.7664, 0.0014)
c_cpixfe (2.8916, 0.0023);
c_emp (4.3766, 0.0000);
c_gdp (1.7828, 0.0182);
c_rtl_sls (3.0025, 0.0010)
3rd
BAX
c_cpixfe (2.8758, 0.0002);
c_emp (3.0365, 0.0016);
c_rtl_sls (2.4350, 0.0035)
R2
us_gdp (1.5311, 0.0001);
us_nfp (3.2642, 0.0000);
us_gdp (1.0231, 0.0058);
us_nfp (2.7792, 0.0000)
2nd
BAX
Significant surprises28,
Post-FAD 1st Half
(Oct.31/00-July 21/04)
0.2251
us_cpixfe (1.4323, 0.0053);
us_gdp (1.4152, 0.0414);
us_hr_earn (1.6987, 0.0268);
us_ind_prod (0.8430,
0.0433);
us_ism (1.5801, 0.0218);
us_nfp (5.0138, 0.0000);
us_ppixfe (0.9790, 0.0108);
us_unemp (-1.2933, 0.0409)
c_cpixfe (1.3067, 0.0255);
c_emp (1.6305, 0.0224);
c_hsg_sts (-0.8984, 0.0231)
10year
Bond
CADUS
excha
nge
rate
us_cpixfe (1.0933, 0.0304);
us_ism (1.2955, 0.0099);
us_nfp (3.8224, 0.0000);
us_ppi (-0.8034, 0.0283);
us_ppixfe (0.9975, 0.0072)
c_emp (0.2264, 0.0000);
c_lead_ind (0.0785, ,
0.0026);
c_manu_ship (0.0870,
0.0395);
c_rtl_sls (0.1672, 0.0007)
us_cpixfe (1.9551, 0.0163);
us_ind_prod (1.8255,
0.0034);
us_ism (2.5423, 0.0327);
us_nfp (5.1961, 0.0000)
0.1579
c_curr_acct (-2.6186,
0.0018);
c_hsg_sts (-1.5156, 0.0063);
c_ivey (2.2252, 0.0084)
us_ppixfe (1.2387, 0.0045)
c_curr_acct (3.3261,
0.0256);
c_emp (2.5218, 0.0000)
0.1692
us_ism (1.8670, 0.0387);
us_nfp (3.8319, 0.0003)
c_emp (0.1716, 0.0108);
c_lead_ind (0.0857, 0.0019);
c_rtl_sls (0.1733, 0.0085)
0.0944
us_ism (1.0035, 0.0186);
us_nfp (4.1649, 0.0000);
us_ppi (-0.8990. 0.0087);
us_ppixfe (1.4020, 0.0031)
c_cpixfe (0.1636, 0.0116);
c_emp (0.3016, 0.0000);
c_manu_ship (0.1671,
0.0353);
c_mrch_trade (0.1315,
0.0245);
c_rtl_sls (0.1614, 0.0357)
us_hr_earn (-0.1204,
0.0206);
us_ppi (0.1135, 0.0148);
us_trd_bal (-0.0941, 0.0355)
us_ppi (0.0717, 0.0496);
us_trd_bal (-0.1123, 0.0025)
47
0.1982
0.1160