The effect of ECB communication
on the government bond market for
eurozone countries
University of Amsterdam
Master Economics – Monetary Policy & Banking
Master Thesis
Toon Eggink
S10662693
Supervisor: prof. dr. dhr. M. Giuliodori dr. dhr. W.E. Romp
Date: 4-Aug-2014 Word Count: 7,708
Abstract
This thesis examines the durable effect of ECB communication on the government bond market for eurozone countries. The dataset related to the Communication Index is uniquely created and the ECB Post-Meeting Comment published by the Financial Markets Research of the Rabobank is used as source. The analysis includes multiple money market securities and government bonds to create an overall assessment of the effect on financial markets. The time period covered in the model ranges from January 2002 up to and including April 2014.
The results show that the ECB can steer interest rates and that long-term bonds are most affected. The main conclusion is that the effect is more pronounced for peripheral eurozone countries and in the period post the Euro-Crisis. The findings have some clear implications for
Table of Contents
1. Introduction ... 1
2. Literature Review ... 3
2.1 – Central bank communication ... 3
2.2 – Research models ... 4 3. Model Specifications ... 6 3.1 – Communication Index ... 6 3.2 – Interest Rates ... 8 3.3 – Control Variables ... 10 3.4 – Hypotheses ... 12 4. Regression Analysis ... 13 4.1 – Framework ... 13 4.2 – Baseline model ... 14
4.3 – Eurozone panel data ... 15
5. Conclusions ... 16
References ... 18
Appendix A: Method of coding the Press Conferences ... 20
Appendix B: Dataset ... 24
Appendix C: Overview of ECB Communication Index ... 25
1. Introduction
This section functions as introduction to the thesis and highlights the contributions of this study to academia. Next to this, the section specifies the research question as well as the structure of the study.
The speech of Mario Draghi on 26-July-2012 has affected financial markets to a large extent and the statements made during the speech have lead to the end of the Euro Crisis. Communication of the ECB has gained increased attention on the government bond market for eurozone countries since that moment. As an example, figure 1 shows the developments on a generic 10-years government bond of the Netherlands and Spain during the day of the monthly Press Conference of the ECB.
Figure 1: Effect of ECB Communication on a Dutch and Spanish bond
Source: Bloomberg. Note: Spread levels vs. Bund for a generic 10-years Spanish and Dutch
government bond in basis points. The date of observation is 8-May-2014 and the Press Conference is coded as -2 (i.e. very dovish).
The figure shows that the Press Conference affects spread levels of a Spanish government bond more strongly than of a similar Dutch bond during the day. Hence, the effect of ECB Communication is more pronounced for government bonds of peripheral eurozone countries then for core eurozone countries. The bonds covered in the figure have a term period of 10-years and it is proven in this study that long-term bonds are most affected by the Press Conference of the ECB. This study focuses on the durable effect of ECB communication and one can observe that the yield levels end lower on the day of the Press Conference. The main conclusion of this thesis is that the impact on spread levels of a 10-years government bond for peripheral eurozone countries is almost four times larger than for core countries in the period after the Euro-Crisis.
The communication events of the ECB are identified by a unique Communication Index. I have coded the monthly Press Conferences of the ECB from January 2002 up to and including April 2014 and made use of the ECB Post-Meeting Comments of the Financial Markets Research Team of the Rabobank as source. The ECB Post-Meeting Comments discuss the Press Conferences from different perspectives. As a consequence, the Communication Index captures both the impact on the eurozone economy as well as on financial markets.
-10
09:00 11:45 14:30 17:15
The research question of this study is ‘what is the durable effect of ECB communication on
the government bond market for eurozone countries?’ The effect is estimated by an empirical
analysis and theory explains via what channels communication might have an effect. The dataset is split by the set of countries and time period in order to evaluate the effect differentials.
The goal is to evaluate the rationale for implementing communication as monetary policy instrument and to identify to what extent financial markets are affected. The study analyzes the impact of ECB communication on the entire yield curves of multiple eurozone countries. The results are of quantitative scale in order to estimate the differences among eurozone countries and time periods.
The approach of this thesis is similar to the study of Musard-Gies (2006) and the durable effect is analyzed by taking the difference in end-of-day yield levels. I have decided to broaden the set of dependent variables and include both money market-securities as well as government bonds for multiple eurozone countries. The dataset contains up-to-date numbers and in particular most recent years are of interest as communication by the ECB has gained attention on financial markets.
The dataset related to ECB Communication is uniquely created based on the ECB Post-Meeting Comments. I coded the policy stance of the Press Conference according to the tone. Appendix A describes the approach in more detail and the methodology is comparable to Jansen & De Haan (2005) and Hubert (2012). The ECB Post-Meeting Comments are checked for certain key phrases and are labeled with a code ranging from -2 to +2. The Communication Index shows a similar trend as the Monetary Policy Communicator-Index of the KOF-Institute. The ECB Post-Meeting Comments are valuable as it highlights the effect of both the Introductory Statement as well the Q&A-Session and focuses on the impact on financial markets.
The conclusions of this thesis provide insight in the role of communication in monetary policy and the impact on financial markets. A policy recommendation for central banks is to adopt a clear communication strategy as the tone of a Press Conference can steer financial markets. In particular peripheral eurozone countries would benefit from a (very) dovish policy stance of the ECB. Next to this, agents on financial markets can anticipate a Press Conference of the ECB and take a trading position in advance of a meeting.
The outline of this paper is as follows. Section 2 provides an overview of related research papers. In section 3, the model is specified and includes a descriptive analysis of the dataset. In section 4, the regression analysis will be executed for multiple samples and the results are discussed. Section 5 provides an overview of the conclusions of this thesis. Moreover, some limitations and policy recommendations are highlighted.
2. Literature Review
The topic of central bank communication has become more relevant in academia during recent years and the topic is analyzed via multiple approaches. This section highlights several papers that relate to the topic and the different research methodologies. Additionally, the section gives insight in the role of communication in monetary policy over time and in the communication strategies of central banks around the world.
The thesis enriches the current literature by three aspects, which are the unique approach to quantify the communication events of the ECB, the type of securities included in the model and the set of countries. The combination of the three aspects makes it possible to execute empirical analyses that provide renewed insights in the effect of ECB communication.
At first, the Communication Index is based on publications of the Rabobank Financial Markets Research team and the source is unique. The Analysts of the Rabobank aim to highlight both economic as well as financial implications of certain ECB meetings. For example, the ECB Post-Meeting Comments highlight the market expectations or the liquidity risk of certain monetary policy instruments and focus also on the inflation figures. Studies like Musard-Gies (2006) and Berger, Ehrmann and Fratzscher (2007) analyze communication events from an economic perspective and filter for economic related key phrases only. Hubert (2012) widens the scope and analyzes the Press Conferences via news headlines as published by Reuters. The Communication Index used in this thesis focuses on how investors are informed about the ECB Press Conference and therefore analyzes the channel of communication in its purest form.
Second, the dataset related to the dependent variable consists of multiple short-term securities and government bonds up to a maturity period of 10-years. Previous studies have focused on either money market instruments, like Bernoth and von Hagen (2006), or government bonds, like Musard-Gies (2006). This thesis aims to give a complete assessment of the effect on fixed income markets and focuses on the entire yield curve.
Lastly, the model covers a broad set of countries in order to analyze the effects on government bonds of specific countries. This dataset in this thesis is split for government bonds of core and peripheral eurozone countries. Beetsma, Giuliodori, de Jong and Widijanto (2013) analyzed the effect of multiple news events on the interest rate spread levels of GIIPS- and non-GIIPS countries and it is concluded that the effect is more pronounced for non-GIIPS countries.
2.1 – Central bank communication
The thesis focuses on the Press Conferences of the President of the Governing Council and neglects other communication events of the ECB. Other studies have focused on different types of central bank communication and Blinder et al. (2008) have provided an extensive summary of related studies. All studies focused on the central question what the optimal level of central bank
transparency is and the main conclusion is that transparency has a strong effect on agent’s expectations via different channels.
Michael Woodford (2001) is one of the first researchers who addressed central bank communication as a monetary policy instrument and argued that each central bank should have a clear communication strategy. He found that communication might be an effective tool in order to manage the public’s expectations. Kohn and Sack (2003) have analyzed the concept of communication more elaborately and evaluated different communication events by the FED. They included data related to the testimonials of the Congress, the speeches of Greenspan and statements of the Federal Open Market Committee (FOMC) for the time period of 1989-2003. The main conclusions of the study is that central bank talk clearly has an effect on interest rates and that the effect is most pronounced on the short- and medium-term part of the curve. Additionally, speeches have a larger effect than FOMC-testimony and influence government bonds with a long-term maturity as well. The study functions as rationale on how to identify central bank communication.
Central bank communication has been studied from different perspectives. De Haan, Amtembrink and Waller (2004) investigate communication from the perspective of a sender (i.e. the central bank) and a receiver (i.e. the media) and conclude that several newswires interpret communication events in a different way. The findings gave rise to new studies like Goodhart (2005) and Ehrmann and Fratzscher (2007) who focused on the effects of different communication events and reporting channels. Hubert (2012) created a qualitative Communication Index of the ECB based on all statements of members of the Governing Council. The dataset includes speeches, interviews and testimonies related to monetary policy for a specific month and the author used the Reuters’ headlines as source.
2.2 – Research models
Researchers have used quantitative as well as qualitative indicators to give an assessment of a communication event. Jansen and De Haan (2005) provided a qualitative assessment of each ECB Press Conference and made use of a coding approach to create the corresponding dataset. The authors labeled each policy statement with a positive, negative or null value according to the tone of the statement. Rosa and Verga (2005) have extended this approach and coded the communication content of ECB Press Conferences on a range of 5 values. Although this coding approach might be subjective, the method has proven to be successful in Musard-Gies (2006), Berger, Ehrmann and Fratzscher (2007) and Hubert (2012). To correct for subjective outcomes, Hubert (2012) tests his dataset to the Monetary Policy Communication-Index as designed by the KOF and this thesis will follow the same line.
I have specifically chosen to focus on qualitative criteria in the coding approach in order to analyze the content of the Press Conferences. Several studies have focused on quantitative criteria. Laver, Benoit & Garry (2003) introduced the idea of wordscores to create a reliable dataset for communication and Berger, de Haan and Sturm (2006) executed a word count for
each section of the Introductory Statement. Hubert (2012) has combined the quantitative and qualitative criteria and focused also on the frequency of statements made during a specific month. The index created by Hubert covers both the intensity as well as the frequency of ECB communication events during the month.
Musard-Gies (2006) introduced another research perspective and the paper focused on the relative tone of ECB Introductory Statements. The author calculated the change in the tone of statements from one observation to another and evaluated the effect on interest rates.
This thesis aims to give a complete assessment of the effect of ECB communication on money markets and the government bond market for eurozone countries. Musard-Gies (2006) has focused on the change in German government bond yields with different maturities and concluded that the medium part of the curve (i.e. 6- and 12-months government bonds) has been affected the most. The study aimed to evaluate whether the effect of communication is durable and hence defined durable as the difference between the end-of-day interest rates minus the interest rate at the day before the ECB Press Conference.
The scope in academia has changed considerably over time and research papers have highlighted different aspects of the communication channel. Early papers focused on the predictability of central bank policy actions. Among others, Lange et al (2003) and Swanson (2006) analyzed the effect on futures and expectations of the interest rates path. More recent, the focus has shifted towards the effect on financial market asset prices like overnight interest rates or the exchange rate. The studies are of interest for this thesis to estimate via what channels communication might have an effect and to create the framework.
For the set of countries included in the analysis, Blinder et al. (2006) described that it is most common in academia to focus on benchmark bonds like the German Bund or other euro area related securities like the Euribor-rates. Gade at al. (2014) evaluated the impact of political communication on different government bonds within the eurozone and concluded that mainly the connotation of statements does impact the yield curve. Beetsma et al. (2013) made a distinction between two sets of countries and split the euro area into GIIPS and non-GIIPS countries.
The hypotheses for this thesis are based on, among other, papers executed by Musard-Gies (2006), Anderson et al. (2006) and Beetsma et al. (2013). Musard-Musard-Gies (2006) showed that ECB communication does affect the medium-part of the yield curve. Andersson et al. (2006) concluded that adjustments in prices are quick and new information of, for example, macroeconomic announcements or ECB monetary policy releases is usually incorporated into prices within five minutes of announcements. Therefore, I expect that the Press Conferences do have a strong effect on financial markets and in particular on bonds with a maturity of 1-year and 2-years. Moreover, Beetsma et al. (2013) found that the effect of news is more pronounced for GIIPS countries and that spillover effects among euro area countries arise.
The literature review shows that central bank communication can be analyzed via multiple approaches. This thesis focuses solely on the Press Conferences of the ECB and follows the model as designed by Musard-Gies. The study of Hubert is helpful to create the Communication Index and touches upon the approach initiated by de Haan et al. The results of this thesis can be tested to other studies like Beetsma et al, Gade et al and Musard-Gies.
3. Model Specifications
In this section, the economic model will be described with reference to general theories and literature. The analysis is descriptive of nature and the findings are supported by figures.
The model has the form of
, where
is the difference in the yield level on a particular security before- and after the
ECB-meeting. More specifically, the dependent variable can be written as which is
the difference of the end-of-day interest rate on a particular security at the day of the ECB-meeting and the end of day interest rate one day before the ECB-ECB-meeting.
represents the communication-index of ECB-speeches and has the form of a code related to the tone of the speech within a range of five values. indicates the change in the main refinancing rate as set by the ECB relative to last month. The refinancing rate is published on the day of the Press Conference at 13.45hour and the change in the nominal interest rate might affect the interest rate term structure. It should be highlighted that these two independent variables are subject to changes per time period only and do not vary per eurozone country. The model includes multiple control variables (i.e. the vector ) that cover for trends related to the economy as well as financial markets.
3.1 – Communication Index
The Communication Index covers the monthly Press Conference of the President of the Governing Council and each Press Conference is coded according to the tone. The Press Conference consists of the Introductory Statement and the corresponding Q&A-session and the period included in the model is January 2002 up to and including April 2014.
The methodology for coding consists of two steps, which are a content analysis of the Introductory Statement and the assessment of the Rabobank Financial Markets Research Team. The content analysis gives insight in the policy actions announced during the meeting and is helpful to analyze the stance of the ECB. The Introductory Statements are filtered for certain key phrases which are listed in appendix A. The classification functions as indicator for the final index. In the second step, the ECB Post-Meeting Comments are analyzed for certain key phrases and common expressions. The Analysts use a standard structure for the ECB Post-Meeting Comments each month and this makes it is possible to filter for key statements. The Post-Meeting Comment starts with a concise summary of the Press Conference and evaluates the economic
figures. Next, the Analysts interpret the current policy stance and highlight the possibilities of future policy actions. The Analysts focus on the tone of the Press Conference, for example hawkish or dovish, and the expectations people had about the Press Conference. Appendix A provides an overview of key statements and the criteria for each code.
The ECB Post-Meeting Comment captures the channel of communication towards financial markets and ensures the uniqueness of the dataset. Additionally, the ECB Post-Meeting Comments are preferred over the Introductory Statement to identify communication by the ECB for three reasons. First, the Post-Meeting Comment discusses the tone of the Press Conference relative to previous Press Conferences and hence highlights the change in the stance of the ECB. A change in the policy stance might have a large effect, as central bankers should take the time-inconsistency problem and credibility of their statements into consideration. Second, the expectations about the Press Conference are highlighted. As a result, the code captures possible surprises to the market. Third, the Analysts discuss the role of communication as policy instruments and evaluates why the President of the Governing Council makes certain statements. Communication has become more important since the Euro Crisis as the threat of the Zero-Lower Bound is more apparent and the ECB implemented new communication strategies.
Blinder et al. (2008) show that communication is an effective tool in order to provide transparency to private agents about the attitude of policymakers towards certain events. Peter Praet (2013) explains the rationale for adopting forward guidance by the ECB according to two components. One the one hand, communication has an ‘Odyssean’ element in which the central bank provides more insights in the drivers (i.e. the reaction function) of monetary policy. On the other hand, there is a ‘Delphic’ element in which the economic outlook is made public. The latter touches upon the role of research within the central bank and Anderson, Dillen and Selin (2006) found that financial markets do react significantly to the macroeconomic figures published by central banks.
Mishkin (2010) provides an overview of how communication by central banks works in practice. He concludes that the ECB makes relatively little use of communication tools in comparison to other central banks. The ECB organizes a monthly Press Conference in which they announce the refinancing rate and provide an assessment of economic as well as monetary developments. The outline of these Press Conferences is identical each month and the Introductory Statement as well as the Q&A-session provides room for forward guidance. In contrast to other central banks, the ECB does not make use of any other formal communication event. Speeches and interviews of ECB-members are neglected in this study in order to ensure the consistency and structure of the coding methodology.
The Communication Index is presented in figure 2 and the dataset covers 143 meetings. The distribution of codes is well balanced among hawkish and dovish observations and the tone of the Press Conference is modest in most observations (i.e. a code of 1 or -1).
The dotted line represents the Monetary Policy Communicator-index created by the KOF and it can be concluded that both indices follow a similar pattern. The unique dataset of the
Communication Index is economic significant. Moreover, the trend of the Communication Index is comparable to the level of the main refinancing rate (which is plotted in the figure 4).
Figure 2: Communication Index and the Monetary Policy Communicator-index from KOF
Source: Dataset (see appendix C) and KOF-Institute. Note: Distribution of observations: Very
Dovish (-2): 16, Dovish (-1): 28, Neutral (0): 47, Hawkish (1): 36, Very Hawkish (2): 16.
3.2 – Interest Rates
The effect of communication by the ECB is analyzed for multiple securities and government bonds. The baseline model of the empirical analysis relates to money market-securities and German Government bonds. In addition to this, the spread levels for government bonds of multiple eurozone countries are used as dependent variable.
The money market-securities included in the model are EONIA, one month-, three months-, six months- and twelve months-Euribor rates. Theory shows that short-term interest rates are highly affected by the level of the ECB main refinancing rate and that the expectations about the path of future interest rates does influence today’s yield levels. Money markets are efficient and any incoming information as announced in the Press Conference is instantly priced in interest rates.
Moreover, the baseline model consists of German government bonds and includes yield levels for a generic six months, twelve months, two years (Schatz), five years (Bobl) and ten years (Bund) bond. Germany functions as benchmark country and is considered as less risky eurozone country on financial markets. In total, ten securities with different maturities are included and hence it is possible to quantify the effect of the Communication Index on the entire yield curve.
The change in the yield levels has the form of and is calculated by the
difference of the end-of-day yield level at the day of the Press Conference and the end-of-day level at the day before the Press Conference. The numbers are denominated in basis points for which holds that 100 basis points is equal to 1 percent.
I have used Bloomberg as source for the levels of all securities and bonds. Bloomberg is widely used by financial agents and provides accurate and up-to-date yield levels. Another benefit is that Bloomberg has created indices related to ‘generic’ securities and therefore the -1.00 -0.50 0.00 0.50 1.00 -3 -2 -1 0 1 2 3 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
dataset is not affected by any day counting effects. In Appendix B, a table consisting of the specifications of all securities and the corresponding Bloomberg-tickers is presented.
In total, 3152 observations are included in the baseline model of this study. The trends in rates are comparable to the nominal interest rate set by the ECB. Next to this, the trends in yield level of bonds with a different term period are similar and are in line with the principal component theory.
Moreover, the model includes the government bonds of ten eurozone countries. The dataset has the form of panel data and five bonds with a different term period are included for each country. The bonds have a maturity period of six months, twelve months, two years, five years and ten years. The numbers are in spread levels relative to the corresponding German government bond and the end-of-day change in spread levels is calculated by the same methodology. The countries are Austria, Belgium, Finland, France, Greece, Ireland, Italy, the Netherlands, Portugal and Spain. The set of countries can be split into two groups and will be labeled as core (Austria, Belgium, Finland, France and the Netherlands) and peripheral eurozone countries (Greece, Ireland, Italy, Portugal and Spain).
Figure 3 represents the spread levels on a 10-years government bond for all eurozone countries and three features arise. First, the time span can be split in distinctive periods that relate to the Financial Crisis of 2008 and the Euro Crisis of 2011 and 2012. Second, eurozone countries show a different performance and the gap between core and peripheral eurozone countries have increased since 2008. Third, the dotted line shows that communication by the ECB matters. The line represents the date of Draghi’s ‘Whatever it takes’-speech and spread levels for peripheral eurozone countries have tightened significantly since then.
Figure 3: Spread levels for a 10-years government bond of multiple eurozone countries
Source: Bloomberg. Note: Spread levels vs. the Bund in basis points
The trends as identified for the 10-years eurozone government bond are also apparent for the bonds with other maturities. Therefore, bonds with a maturity of 6 months-, 12 months- 2years and 5 years face a similar pattern.
0 1000 2000 3000
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Austria Belgium Finland France Greece
3.3 – Control Variables
The model includes five control variables, which are the ECB main refinancing rate, the VIX-index, the US-Treasury Bill, the euro dollar exchange rate and yield levels for the Bund.
The model is corrected for developments in monetary policy of the ECB. A variable, identified as in the general formula, is included to cover the change in the main refinancing rate. Moreover, a dummy variable is included to control for announcement of (unconventional) monetary policy instruments. Such announcements have a direct effect on interest rates rather than the signaling functioning of communication. Hence, these observations fall beyond the scope of this analysis.
Figure 4 shows the main refinancing rate of the ECB over time and both an upward as well as downward trend can be observed. The dataset covers 143 observations and the refinancing rate has fluctuated within the range of 0.25% and 4.25%. The Financial Crisis of 2008 marks the turning point in monetary policy and as a consequence the interest rate has been lowered drastically.
Figure 4: ECB Main Refinancing Rate
Source: www.ecb.europa.eu. Note: Rates published in percentages.
The study focuses on day-to-day changes of government bond yields and the ECB has changed the nominal interest rate 25 times during the sample period. 11 of these observations are related to an increase of the interest rate while the interest rate has been cut 14 times. The ECB changed the interest rate by 25 basis points in most cases.1
In the period after May 2009, the ECB implemented Long-Term Refinancing Operations (LTRO) and amended the collateral framework for financial institutions. The policy instruments have an impact on the yield levels by the liquidity effect. The asterisks in figure 4 indicate the meetings in which the ECB announced to implement (unconventional) monetary instruments. The dataset consists of 132 observations after excluding these observations.
1 The collapse of Lehmann Brothers in September 2008 gave rise to an extreme loose monetary policy and the ECB decided to lower the interest rate by respectively 50 basis points in November and 75 basis points in December 2008. Moreover, the ECB organized an extra meeting in October 2008 in which it announced to decrease the main refinancing rate by 50 basis points.
0 1 2 3 4 5 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
A vector of control variables is included and one can distinguish four variables. At first, a variable that covers for market sentiment is included to correct for certain trends on financial markets. The Chicago Board Options Exchange Volatility Index functions a proxy for market sentiment and estimates the expected volatility in global financial markets. I prefer the VIX-index to the iTraxx-index, another index that is used to capture market sentiment, since it provides an overview of sentiments among all global markets rather than European markets only.
Moreover, the interest rate for the 10-year US-Treasury Bill is included as control variable. The US Treasury-bill is an indicator of global economic trends and covers for macroeconomic developments beyond the scope of the eurozone. Additionally, the variable corrects for possible substitution effects between eurozone- and US-government bonds.
Third, the levels for the euro-dollar exchange rate are included. The exchange rate controls for regional aggregate risk related to both economic as well as financial aspects. The rationale for including this variable is that more uncertainty in, for example, the eurozone drives investors out of Europe and weakens the value of the euro relative to the dollar.
Finally, yield levels for the 10-year German government bond are used as control variable. The variable captures some regional market sentiments that might arise and the effect of ECB communication is analyzed more precisely. The control variable is only included in regressions where the securities related to Germany are not included. Although the spread levels on government bonds will be computed as the difference relative to the related German government bond, the control variable will still be included to cover for market sentiment and co-movements of bonds.
The period included in the model is characterized by a strong economic expansion from 2002 up to 2008 and a severe global recession afterwards. The yield levels for US Treasury-bills are included to cover for global economic business cycles and the VIX-index represents the expected future volatility. The index is useful to control for others factors apparent on global financial markets and the figure below shows three periods of unrest on financial markets.2 Figure 5: VIX-Index and Asset Substitution
Source: Bloomberg. Note: VIX-Index in points and yields levels for a 10-years German and Spanish
government bond listed as percentage.
2 In line with the trends on the government bond market, the VIX-index increased after the Financial Crisis of 2008 and the Euro Crisis of 2011 and 2012.
0 5 10 0 50 100 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 VIX-index (left-scale) Spain (right-scale) Germany (right-scale)
Next to this, the right-scale axis of figure 5 provides insight in asset substitution in the eurozone government bond market. For example, a peak in the volatility index as can be observed in November 2011 indicates that market sentiment worsened. In line with the risk-off modus of investors, prices of lower-rated financial instruments like Spanish government bonds fell and the corresponding interest rates increased. Investors look for ‘safe haven’ investment and German Bunds are becoming attractive reflected by lower interest rates.
The main conclusion of this section is that the model holds in practice. The findings show that financial markets are affected by communication and the Communication Index proves to be economic significant. The time span of the dataset covers both a period of economic expansion as well as a severe recession. The developments over time and among (groups of) countries function as rationale for introducing certain splits in the empirical analysis.
3.4 – Hypotheses
The hypotheses of this thesis are based on the literature review as well as the model specifications. The hypothesis can be split up in three components.
At first, I expect that ECB communication has an impact on financial markets and hence the tone of the Press Conference does steer interest rates in the desired direction. This implies that the Governing Council can lower interest rates by making use of a (very) dovish tone in its statements and therefore a positive coefficient is expected in the regression analysis.
Second, I expect that the effect on long-term bonds is larger than on short-term securities. The rationale behind this is that trading in short-term securities is more efficient and hence the future policy stance of the ECB is already priced in. Long-term bonds however are subject to multiple factors and the tone of communication could dominate those forces. In line with this, my hypothesis is that government bonds from peripheral eurozone countries will be more affected by the tone of Press Conference then core eurozone countries. The concept of asset substitution does have a large effect and for example a very dovish tone would make investors more willing to take risk. Investors will switch out of safe-haven securities into high-yield assets like peripheral eurozone government bonds. Hence the coefficient of the regression analysis for this specific sub-sample will be larger in comparison to those of for example Germany and other core eurozone countries.
Finally, I expect that changes in the tone of the Introductory Statements from one month to another would have a larger effect than the actual tone. Moreover, I think that an extreme tone of the Press Conference, which implies either a very dovish or very hawkish tone, has a larger effect than a modest tone. The rationale for this hypothesis relates to the time-inconsistency problem of central bank communication and hence central bankers use a prudent communication strategy.
4. Regression Analysis
The empirical analysis of this study consists of three parts, which are the regressions related to the baseline-model for money market securities, for German government bond and to the eurozone panel dataset. The section starts by explaining the framework of the regressions and the statistical implications. Next, I focus on the most relevant results. One can consult Appendix D for an overview of all regression results (i.e. the coefficient estimator and the standard errors) and its statistical inferences.
The main finding is that the Communication Index has a large and significant effect on long-term bonds of peripheral eurozone countries in the period after the Euro Crisis. The coefficient estimator is four times larger for peripheral then for core eurozone countries. Moreover, I found that asset substitution is apparent in the eurozone government bond market. The effect on money market-securities is muted and German Government bonds are most affected in the medium-part of the curve.
4.1 – Framework
The estimation technique is standard OLS and the regressions are executed in robust format to correct for heteroskedasticity.3 The regression method is identical to the study of Gade et al. (2013). This method is preferred since it captures the effect of individual programme countries as well as the possible contagion effects. The focus will be on t-values and the corresponding p-statistics to analyze the significance of each variable. The data are corrected for possible unit roots and serial correlation in fixed time c.q. country effects by taking the first difference of the rates.
The variables have been tested for collinearity and a correlation matrix of the variables can be found in Appendix D. The matrix shows that the Communication Index is related to the change in the ECB main refinancing rate and that the bonds from a particular country with different term periods are correlated. I found that none of the variables are a linear combination of another independent variable and hence the model is not violated by multi collinearity.4
In case of eurozone panel data regressions, the effect of communication on interest rate changes is linear as is shown by a simple scatter plot for each regression. The effect seems to be non-linear for the baseline regressions and communication seems to have a higher effect in case of extreme values in the Communication Index (i.e. a code of -2/2).
3 The baseline- as well as the panel data regressions has been tested for homoscedasticity according the Breusch-Pagan Cook Weisberg test. It is concluded that heteroskedasticity might be apparent.
4 The model is checked for multi collinearity according to the method of variance inflation factor (VIF). As a rule of thumb, I have assumed that a VIF-value larger than 10 implies collinearity. The VIF-estimators for all variables are lower than the threshold and the mean VIF fluctuates around 1.5 and 2.5 in most regression.
4.2 – Baseline model
The regression results related to the baseline model are listed in Table 5 and 6. A one-unit increase in the Communication Index makes the 3-months and 6-months Euribor rates increases by respectively 0.13 and 0.10 basis points.5 The results prove that the model is economic significant in both bases and hence the F-statistics does hold in all regressions.
The level of the Communication Index fails to explain trends underlying the change in German government bonds and other money market-securities. The findings do not provide any strong conclusions and I amend the independent variable into the change of the Communication Index instead of the actual level.
A change in the Communication Index makes the yield levels for the Bobl increase with 1.02 basis points as listed in column 4 of Table 8. Changes in the Communication Index have a positive and significant effect on both short-dated Euribor rates as well as on different spots of the curve for German government bonds. The effect is most pronounced in the medium-term part of the German yield curve. The model proves to have explanatory power, as the F-statistic is significant at a 5% level.
The results of the sub-regressions, which are summarized in Table 10, show that the effect of changes in the Communication Index has become more pronounced on long-term German government bonds in the period after 2009. Row 5 shows that the coefficient estimator for the 2-, 5- and 10 years-German government bond is respectively 1.43, 1.90 and 1.36. The observations in which unconventional policy instruments were announced are omitted. If we include these observations, then the coefficient estimator is larger. Hence, this proves that announcements related to unconventional policy instruments do influence the change in rates.
Moreover, the findings suggest that the effect of communication by the ECB has become more pronounced in meetings after Draghi’s ‘Whatever it takes’-speech. The sample contains 17 observations and row 8 of Table 10 shows that a one unit increase in the Communication Index makes spread levels change by 3.68 basis points.
The main conclusion is that the effect of ECB Communication is muted in the baseline model. The change in the tone of the Press Conference does a better job in explaining the trends then the actual level. The coefficient estimators provide limited insights that make it hard to make any clear policy recommendations based on the results. Moreover, the research model has explanatory power and the variables prove to be significant. The split per time periods provide interesting insights and will be used in further analysis. Moreover, the announcements of unconventional monetary policy instruments do have a significant effect and therefore need to be omitted. Finally, one should evaluate whether the level or the change in the Communication Index steer interest rates in further analysis.
5 The coefficient for the 3-months Euribor rate is 0.13, standard error 0.0041 and is significant at a 1%-level. The coefficient for the 3-months Euribor rate is 0.10, standard error 0.049 and is significant at a 5%-level. In further regression results, one can consult Appendix D for the level of the coefficient estimator, standard errors and significance level.
4.3 – Eurozone panel data
The regressions including panel data for eurozone countries are executed via a comparable method as the baseline regressions. The dataset consists of daily observations for spread levels relative to German governments bonds and day-to-day changes in the Bund are included as control variable. In line with the dataset before, 143 observations for ECB communication are included and the same dummy variables will be applied.
The results in Table 12 show that ECB communication has a positive and significant effect on spread levels for bonds with a maturity of 2years, 5years and 10years. The effect is most pronounced for bonds with a maturity of 2years and the corresponding coefficient estimator is equal to 12.4.6 The effect of changes in the Communication Index is not significant for any bond and the analysis continues to focus on the actual level in further regressions.
The findings for several sub-samples are presented in Table 14 and three conclusions can be made. First, a clear difference between core and peripheral eurozone countries arises. Spread levels for core countries change by 5.52 basis points and by 19.90 basis points for peripheral countries in case of a change in the Communication Index.
Second, the impact differential is larger in the period after the Euro Crisis. Row 8 and 10 of Table 14 represent the coefficient estimator for the curve of the two country sets. The effect on core eurozone spread levels is muted while peripheral’s spread levels increase by respectively 200, 177 and 154 basis points for a 2years, 5years and 10years bond.7
Third, a non-linear effect arises for the 10-years government bond and the effect is comparable to the sample that covers all observations. Next to this, the level of significance is higher if including -2/2-codes only.
I have also analyzed the effect on (changes in) yield levels of eurozone countries and the results are listed in Table 15. The difference with the former regressions is that the explained variable is not corrected for changes in the Bund.
The findings show that the level of communication has a positive and significant impact on the yield levels. Moreover, a clear difference between countries can be observed. The coefficient for a 10y-bond for peripheral countries is equal to 23 basis points and the effect on yield levels is larger relative to core eurozone countries, i.e. a coefficient estimator of 14 basis points.
The results give insight in the substitution effect apparent in the euro government bond market, as the sign of the coefficient estimator is different for the two sets of countries. Yield levels for a 2- and 5-years bond for core eurozone countries decrease after a hawkish speech while the coefficient for peripheral eurozone countries is positive. Row 8 and 10 of Table 15
6 The results are statistically significant at a 1% level and the 10-years bond is sgnificant at a 0.1% level. The results are changed slightly if the sample is corrected for announcements of unconventional monetary policy tools as listed in row 1 of table 14.
7 The coefficients for the sample including core eurozone countries are not significant at any level and hence no clear conclusions can be made. However, if we compare the coefficients of the sample containing peripheral eurozone countries only to the general eurozone regressions, then the large impact on peripheral spread levels becomes apparent.
present the coefficient estimators and the effect differential is most extreme for the 2-years bond. The coefficient for core EZ-countries is -1.14 and for peripheral EZ-countries is 1.79. 8
In general, the conclusions from analyzing the change in yield levels are comparable to the analysis of spread levels.9
The conclusions from the regression analysis show that communication by the ECB does have a strong effect on financial markets. The baseline model is helpful to analyze the (joint-) significance of the research model but the coefficient estimators only show limited impact. The results from the eurozone panel data-regression quantify the effect for different samples and provide clear policy implications.
5. Conclusions
This thesis examines the durable effect of ECB communication on the eurozone bond market. The Press Conferences from January 2002 up to and including April 2014 are coded based on the Post ECB-Meeting Comments of the Rabobank Financial Markets Research. The results from the empirical analysis show three major conclusions.
First, the sign of the coefficient estimator is positive. This indicates that the ECB can steer interest rates in the desired direction. The conclusion holds for both money market securities as well as eurozone government bonds. For changes in yield levels, asset substitution is apparent and a negative relation occurs for core eurozone countries.
Second, the effect is larger for long-term bonds. The coefficient estimator is largest on the 2-years bond and the level is comparable to the 5- and 10-years bond. Moreover, the empirical analysis shows that changes in the Communication Index drive the money market-securities and German government bonds. The level of the Communication Index affects eurozone government bonds.
Third, the effect of communication clearly differs per time period and the set of countries. The effect of ECB communication on financial markets is more pronounced since the Euro Crisis and the effect is higher for peripheral eurozone countries. The change in spread levels of peripheral eurozone countries is almost four times as large as changes in core eurozone countries.
The conclusions of this thesis provide new insights in the effect of ECB communication and some results are unique in comparison to other research papers. This thesis suggests that
8 The coefficient estimator for core EZ-countries has a standard error of 0.21 and is significant at a 0.1%
level. The coefficient estimator for peripheral EZ-countries has a standard error of 0.62 and is significant a 1% level.
9 I have executed multiple regressions for specific countries, as identical to the German government bonds
the long-term part of the curve is most affected. Additionally, it shows the impact differentials between the pre- and post Euro-Crisis samples.
The results of this thesis support the findings of Musard-Gies (2006) as it is shown that communication of the ECB has a positive effect on financial markets. Moreover, it is found that the change in the Communication Index does affect German government bonds. The results related to the divergence among eurozone countries are similar to the findings by Gade et al (2013). In addition to this, the thesis provides unique insight as it concludes that a substitution effect occurs between core and peripheral eurozone countries.
The findings of this thesis have some policy implications. With regard to monetary policy, the ECB should carefully consider whether to repeat certain policy phrases or to provide additional information about the stance. The level, as well as the change of the Communication Index, affects financial markets, but to a different extent. Moreover, the ECB can provide (implicit) support to certain countries as specific components of risk can be targeted by speech.
Traders on financial markets should anticipate communication of the ECB by taking a position as it is shown that ECB Press Conferences have a large market impact. Next to government bonds, the ECB affects also other financial variables like the exchange rate, VIX-index or equity indices.
For research, the thesis shows that the definition of communication affects the scope of the analysis and the dataset. The dataset of this thesis captures a concrete channel of communication and a unique source is used. Hence, the topic is analyzed from the perspective of agents on financial markets. Moreover, the ECB Post-Meeting Comments show that the Q&A-session has a large effect on the actual tone of the Press Conference and should be incorporated in the definition of ECB communication as well.
The empirical analysis faces some limitations. The method of quantifying the tone of communication is ‘subjective by nature’, although I have made us of a strict methodology. The dataset is tested relative to several indicators to ensure the reliability and the Communication Index has proven to be economic significant.
Another limitation is that the Communication Index consists of observations related to the monthly Press Conference only. The model does not capture communication events that take place during a month.
Finally, the regressions are executed in OLS-format and the effect among bonds on the yield curve is neglected. A Principal Component Analysis can estimate such effects. A drawback of a PCA is that it is not possible to analyze bond-specific effects.
References
o
Andersson, M., Dillen, H. & Sellin, P. (2004), ‘Monetary Policy Signaling and Movements inthe Swedish Term Structure of Interest Rates‘, Sveriges Riksbank, Working Paper Series No.
132
o Andersson, M., Hansen, L J. & Sebestyen, S. (2006), ‘Which News Moves the Euro Area Bond
Market?’ European Central Bank, Working Paper Series no. 631
o Armstrong, N. & Mnyanda, L. (2014), ‘European Bonds Jump as Draghi Signals Potential for
June Action’, Bloomberg News Article published on 8-May-2014 17:54:17
o Beetsma, R., Giuliodori, M., de Jong, F., & Widijanto, D. (2013), ‘Spread the News: the
Impact of News on the European Sovereign Bond Markets during the Crisis’, Journal of
International Money and Finance
o Berger, H., de Haan, J. & Sturm, J-E. (2006),‘Does Money Matter in the ECB Strategy: New
Evidence based on ECB Communication, CESifo, Working Paper no. 1652
o Berger, H., Ehrmann, M. & Fratzscher, M. (2006),‘Monetary Policy in the Media’, European Central Bank, Working Paper Series no. 679
o Berk, J. & DeMarzo, P. (2007), ‘ Corporate Finance’, Pearson International Edition
o Blinder, A. (2004), ‘The Quiet Revolution: Central Banking Goes Modern’, Yale University Press
o Blinder, A., Ehrmann, M. Fratzcher,M., de Haan, J. & Jansen, D-J. (2008), ‘Central Bank
Communication and Monetary Policy: A Survey of Theory and Evidence, NBER Working Papers
13932, National Bureau of Economic Research
o Borio, C. & Disyatat, P. (2009),‘Unconventional monetary policies: an appraisal, Bank for International Settlements, BIS Working Papers No. 292
o Brand, C., Buncic, D. & Turunen, J. (2006), ‘The Impact of ECB Monetary Policy Decisions on
Communication on the Yield Curve’, European Central Bank, Working Paper Series no. 657
o Cochrane, J.H (2005), ‘Writing Tips for Ph. D. Students’, Graduate School of Business University of Chicago
o Committee of European Securities Regulators (2010), ‘CESR’s Guidelines on a common definition of European money market funds, CESR/10-049
o De Grauwe, P. & Ji, Y. (2012),‘Mispricing of Sovereign Risk and Multiple Equilibria in the
Eurozone’, CEPS No. 361
o De Haan, J., Amtembrink, F. & Waller, S. (2004), ‘The Transparency and Credibility of the
European Central Bank’, Journal of Common Market Studies
o Draghi, M. (2014), ‘Euro area economic outlook, the ECB’s monetary policy and current policy
challenges’, Statement by Mario Draghi, President of the ECB, prepared for the twenty-ninth
meeting of the International Monetary and Financial Committee, Washington D.C., 10 April 2014
o Draghi, M. (2014), ‘Verbatim of the Remark made by Mario Draghi, Speech by Mario Draghi, President of the ECB, at the Global Investment Conference, London, 26 July 2012
o Ehrmann, M. & Fratzscher, M. (2007), ‘Explaining Monetary Policy in Press Conferences’, European Central Bank, Working Paper Series no. 767
o European Central Bank (2011), ‘The Monetary Policy of the ECB’, European Central Bank o European Commission, ‘The euro’, Economic and Financial Affairs,
http://ec.europa.eu/economy_finance/euro/index_en.htm
o Filardo, A. & Hofmann, B. (2014),‘Forward Guidance at the Zero Lower Bound’, BIS Quarterly Review March 2014
o Gade, T., Salines, M., Glockler, G., & Strodthoff, S. (2013), ‘Loose Lips Sinking Markets? The
Impact of Political Communication on Sovereign Bond Spreads’, European Central Bank,
Working Paper Series no. 150
o Gerlach, S. (2007), ‘Interest Rate Settings by the ECB 1999-2006: Words and Deeds’, International Journal of Central Banking
o Goodhart, J. (2005), ‘Dear Jean-Claude’, Open letter published on Centralbanking.com,
http://www.centralbanking.com/central-banking-journal/feature/2069484/dear-jean-claude
o Hubert, P. (2012), ‘Disentangling Qualitative and Quantitative Central Bank Influence’, OFCE - SciencesPo
o Jansen, D-J. & de Haan, J. (2005), ‘Talking Heads: The Effects of ECB Statements on the
Euro-Dollar Exchange Rate’, Journal of International Money and Finance
o KOF Swiss Economic Institute - http://www.kof.ethz.ch/en/
o Kohn, D. & Sack, B. (2003), ‘Central Bank Talk: Does It Matter and Why?’ Board of Governors of the Federal Reserve System
o Krouse, S. (2012), ‘Investing in PIIGS: Portugal’, Financial News
o Krugman, P. (1998), ‘It’s Baaack: Japan’s Slump and the Return of the Liquidity Traps’, Massachusetts Institute of Technology, Brookings Papers on Economic Activity, 2:1998 o Lange, J., Sack, B. & Whitesell, W. (2003),‘Anticipations of Monetary Policy in Financial
Markets’, Journal of Money, Credit and Banking
o
Laver, M., Benoit, K. & Garry, J. (2003), ‘Extracting Policy Positions from Political Texts UsingWords as Data?’ American Political Science Association, The American Political Science
Review Vol. 97 No. 2
o Mishkin, F. (2010), ‘ The Economics of Money, Banking and Financial Markets’, Columbia Unisversity, Pearson International Edition Ninth Edition
o Musard-Gies, M. (2006), ‘Do European Central Bank’s Statements Steer Interest Rates in the
Euro Zone?’ The Manchester School
o Pattipeilohy, C., van den End, J-W., Tabbae, M., Frost, J., & de Haan, J. (2013), ‘Unconventional Monetary Policy of the ECB during the Financial Crisis: An Assessment and
New Evidence’, DNB Working Paper, Working Paper Series No. 381
o Praet, P. (2013), ‘Forward Guidance and the European Central Bank’, Column published on VoxEU.org, accessed via http://www.bis.org/review/r130807b.pdf
o Riecher, S. & Buergin, R. (2014),‘Draghi’s Patience Snaps on Euro as ECB Threatens
Response’, Bloomberg News Article published on 14-April-2014
o Rosa, C. & Verga, A. (2004), ‘On the Consistency and Effectiveness of Central Bank
Communication: Evidence from the ECB, European Journal of Political Economy
o Stock, J. & Watson, M. (2011), ‘Introduction to Econometrics’, Pearson Education Third Edition
o Swanson, E. (2006),‘Federal Reserve Transparency and Financial Market Forecasts of
Short-Term Interest Rates’, Journal of Money, Credit and Banking
o Walsh, C.E. (2010), ‘Monetary Theory and Policy’, Massachusetts Institute of Technology, Third Edition
o Woodford, M. (2001), ‘Monetary Policy in the Information Economy’, NBER Working Papers 8674, National Bureau of Economic Research
Appendix A: Method of coding the Press Conferences
The dataset of this study is created by a method of coding identical to Jansen & de Haan (2005) and Rosa & Verga (2005). The dataset is unique since it analyzes how financial markets absorb information from the ECB. News agencies like Bloomberg and Reuters provide up-to-date headlines about statements done during the Press Conference and analysts report directly about the Press Conference.
The source of the dataset is the ECB Post-Meeting Comment published by the Financial Markets Research-team of the Rabobank. The analysts of the Rabobank focus on the Press Conference of the President of the Governing Council and evaluate in what way Bloomberg reports about the Press Conference. Therefore, the ECB Post-Meeting Comment provides a mixture of exact statements done during the Press Conference as well as the implications for financial markets.
The ECB Post-Meeting Comments aim to provide a concise summary of the meeting and have a similar structure each month. The Comment starts by highlighting the policy decision of the ECB and discusses the policy stance. Second, the authors evaluate the economic projections as published during the Introductory Statement in more detail. Finally, the analyst interprets the implications for financial markets and makes some predictions for future monetary policy actions.
The document is sent out to clients of the Rabobank and other interested parties. The document is published within a few hours after the meeting. The authors of the document are professionals and have multiple years of experience as analyst/researcher. Additionally, the authors have specialized themselves in a certain geographical- or economic area, in this case the economy of the eurozone, in order to ensure the quality of each research. Therefore, the Comments are reliable and possible (subjective) interpretations of the authors reflect the vision of an expert.
I have codified each monthly meeting and the ECB Post-Meeting Comment function as main source. In order to ensure the objectivity of the coding, I start with a content analysis of the Introductory Statements to get a first impression of the tone of the Press Conference.
The structure of the Introductory Statements is similar each month and I have consulted the first two sections of the Introductory Statements in particular. These sections provide a concise overview of the policy stance of the ECB and the rationale for certain policy decisions. The further part of the Introductory Statements relate to the monetary-, economic- and fiscal figures underlying the policy stance of the ECB’s.
In line with the coding methods used in other studies, I have identified several key phrases of (the first two sections of) the Introductory Statements. The phrases function as signaling words and the table below gives an overview of the key phrases.
Introductory Statement ECB
Very Dovish Plus, unconventional; Key statement
Dovish Accommodative; Prolonged/Extended/Long Period; Present or Lower Int Rate; Use All Instruments
Neutral No announcement of future path; monitoring general developments; waiting for incoming info
Hawkish Inflationary pressure; Temporary nature of instrument; Focus on ECB mandate
Very Hawkish Plus, strong vigilance; Not use all instruments
Based on these phrases, I come up with a first estimation of the tone of the Press Conference. A certain classification is given if the corresponding criteria are met. For example, if the President states “accommodative” and “interest rates will remain present or lower for a
prolonged period of time”, then the Statement is labeled as ‘dovish’. If in addition to this, the Press
Conference includes phrases related to unconventional measures or includes a very dovish statement, then ‘very dovish’ is applied. The same methodology holds vice versa.
It should be noted that the classifications in the content analysis function as indicator for the final index, which will be created via Rabobank Post-ECB Meetings Comments.
We will focus on the Press Conference of 3 April 2014 as example of how the method of coding works in the first stage. The Press Conference of 3 April was very dovish since certain key expressions were present. Mario Draghi stated in the Introductory Statement that ‘we will
monitor developments very closely and will consider all instruments available to us. We are resolute in our determination to maintain a high degree of monetary accommodation and to act swiftly if required. Hence, we do not exclude further monetary policy easing and we firmly reiterate that we continue to expect the key ECB interest rates to remain at present or lower levels for an extended
period of time’
.
Additionally, he emphasized the very dovish tone of speech by means of thephrase ‘The Governing Council is unanimous in its commitment to using also unconventional
instruments within its mandate in order to cope effectively with risks of a too prolonged period of low inflation.’
The classification that would best fit this meeting is very dovish. The first quote explained the accommodative stance of the ECB and the additional sentence related to using unconventional instrument amplified the intentions of the ECB.
The ECB meeting of 7 April 2011 was very hawkish and multiple signaling words justify this classification. After the announcement of an increase in the interest rate, Jean-Claude Trichet highlighted that ‘the adjustment of the current very accommodative monetary policy stance is
warranted in the light of upside risks to price stability that we have identified in our economic analysis.’ Moreover, the sentences ‘we will continue to monitor very closely all developments with respect to upside risks to price stability’ and ‘all the non-standard measures taken during the period of acute financial market tensions are, by construction, temporary in nature’ reveals that future
tightening of monetary policy is not excluded.
In the process of labeling each Press Conference, the content analysis of the Introductory Statement function as first indicator. In the next step, I have consulted the ECB Post-Meeting Comments of the Rabobank Financial Markets Research-team. The Post-Meeting Comments capture the channel of communication towards financial markets and ensure the quality and uniqueness of the dataset. For the Post-Meeting Comments, I focus on different signaling words and some of the key phrases are listed below.
ECB Post-Meeting Comment
-2 Plus, supportive key message; Positive surprise to the market
-1 Dovish; Forward guidance; Ease/loose/accommodative monetary policy
0 Vague message; Expected by markets
1 Hawkish; Inflationary pressure; Closely monitoring
2 Plus, negative key message; Vigilance; Negative surprise to the market
Note that some codes might be ‘overruled’ by certain very strong statements written in the Comment
The same methodology is applied and a code of -2 or 2 can only be given in case the criteria for both categories are met. The ECB Post-Meeting Comments might include certain statements that deviate from the criteria and clearly indicate the tone of the Speech. If so, such phrases can overrule the methodology of coding and are decisive in attaining a certain code to the meeting. For example, the Rabo Comment of June-2008 include the sentence “The regular
hawkish comments were priced in, but Trichet took it a step further, stressing the upside risks to price stability had increased and noting the ECB is now in a “state of heightened alertness”. In such
case, the Press Conference contains a very hawkish signal to the market and it is justified to overrule the regular procedure. Hence, the Press Conference is labeled as ‘very hawkish’.
It is expected that the classifications of both the Introduction Statements as well as the ECB Post-Meeting Comments touch upon identical components. However, three major differences occur between the content analysis of the Introductory Statement and the Post-Meeting Comments.
First, the Post-Meeting Comments specify changes in the tone of the ECB relative to the last meeting. Since the structure of the Introductory Statements is quite similar each meeting, a change in the text or including an extra sentence to emphasize the stance of the ECB is valuable. Second, the Post-Meeting Comment includes the expectations of financial agents about the Press Conference. If for example multiple indicators give rise to a very dovish tone of the ECB and the actual tone is muted, then financial markets perceive the Press Conference as hawkish. Third, the Meeting Comments put the role of communication by the ECB into perspective. The
