Can you change the world by changing your words? : the effect of ECB communication during a financial crisis

Can you change the world by changing your

words? The effect of ECB communication

during a financial crisis.

Bachelor's Thesis, academic year 2014-2015 Faculty of Economics and Business

Universiteit van Amsterdam Christina Haseth, 10141936 Supervisor: Ms L.A. Górnicka

Statement of Originality

This document is written by Student Christina Haseth who

declares to take full responsibility for the contents of this

document.

I declare that the text and the work presented in this

document is original and that no sources other than those

mentioned in the text and its references have been used in

creating it.

The Faculty of Economics and Business is responsible solely

for the supervision of completion of the work, not for the

contents.

Abstract: This thesis investigates whether ECB communication becomes more

important during a time of crisis. The changes in Euribor rates are used as the

dependent variable and each ECB statement from 2004 until 2014 is quantified by

receiving a code of -2, -1, 0 , 1 or 2. The empirical analysis employs multiple ordinary

least squares, with four independent variables with two being control variables. Results

suggest that ECB communication indeed becomes more important during turbulent

times, but the model used in this thesis is a bit uncertain and thus further research

needs to be done.

1

Introduction 2

1. Literature review 4

1.1 The importance of transparency and communication 4

1.2 Types of communication 5

1.3 Empirical evidence 7

1.4 Problems with communication 10

2. Dataset 11

2.1 Communication index 11

2.2 The remaining variables 14

2.3 Data description 16

2.4 The regression: Ordinary least squares 17

3. Empirical analysis 19

3.1 Results from using 12 month Euribor rates 19

3.2 Results from using 1 month Euribor rates 21

3.3 Results from using 1 week Euribor rates 22

3.4 OLS assumptions 22

3.5 Robustness check 23

Conclusion 27

2

Introduction

August 2007 marked the first phase of the global financial crisis, with the initiation of liquidity operations by the European Central Bank (ECB) (P. Lane 2012, 49-67). According to Lane (2012, 49-67) the global financial crisis entered a more intense phase when Lehman Brothers collapsed in September 2008, affecting both Europe as much as the United States in late 2008 and early 2009. This global financial crisis later developed into a sovereign debt crisis (M. Arghyrou, A. Kontonikas, 2012, 658-677).The sovereign crisis caused European policy makers to take extreme measures to limit the negative consequences for the affected countries and prevent it from further spreading (M. Arghyrou, A. Kontonikas, 2012, 658-677).

“Extraordinary times call for extraordinary measures. Responding to the very difficult economic and financial challenges we face, the Federal Reserve has gone beyond traditional monetary policy making to develop new policy tools to address the dysfunctions in the nation's credit markets.” – B. Bernanke, National Press club Luncheon, 2009.

These extraordinary measures are also known as unconventional monetary policy tools and are formed by three elements: 1. large scale liquidity support to banks, 2. forward guidance of ultra-low policy rates over extended policy horizons, 3. large-scale financial market

interventions, in particular huge asset purchases (C. Pattipeilohy, J.W. Van den End, M.

Tabbae, J. Frost and J. De Haan, 2013). These measures are implemented to regain control on the economy when the effect of the crisis on the real economy is large and interest rates are at the zero lower bound. This thesis focuses on the second tool known as forward guidance. Forward guidance aims at clarifying the central banks’ future policy path. For example by signaling low policy rates in the future and thus providing additional stimulus at the zero lower bound and reducing uncertainty (A. Filardo and B. Hoffman, 2014, 37-49). The European Central bank is known to practice another form of forward guidance using indirect signals, often in the form of code words like “vigilance” to provide some sort of idea of future policy decisions (David-Jan Jansen and J. De Haan, 2007). Other central banks like the Federal Reserve (FED) are more explicit by issuing statements with a forward-looking assessment of future monetary policy (Blinder, Ehrmann, Fratzscher, De Haan, Jansen, 2008).

Two channels of transmission are identified, the first is the signaling channel, which enables the central bank to use communication to restore confidence in the markets and influence private expectations about future policy decisions. This channel is especially useful to provide further stimulus to the economy when official interest rates reach the zero lower bound

3

(M. Cecioni, G. Ferrero and A.Secchi, 2011). The other channel of transmission explained by Cecioni et al (2011) is known as the portfolio-balance channel. This mechanism works as follows, central banks purchase public and private securities increasing liquidity and pushing asset prices up, affecting financial conditions and ensuring rebalancing of portfolios (M. Cecioni, G. Ferrero and A.Secchi, 2011).

Central bank communication has become a key policy instrument according to Jansen and De Haan (2005), they argue that the main benefit is the opportunity to directly influence private sector expectations. Investors focus on what central bankers say as these statements are an important indicator of future policy decisions. Their words may well be as important as their actions (D. Jansen and J. De Haan, 2005). Another reason why communication is

important is the degree of transparency that it brings, making monetary policy more credible and effective. It also makes it easier for the public to hold the ECB accountable for its policy

decisions (ECB website). Previous research by inter alia Musard-Gies (2005), Rosa and Verga (2007, 146–175) and Reeves and Sawicki (2007, 207–227) shows that central bank

communication does indeed influence policy rates.

During the recent crisis central banks have been focusing a lot on unconventional monetary policy tools like forward guidance, making it crucial to investigate whether or not this tool is more helpful during these turbulent times. If it is not, finding other monetary policy tools to stimulate the economy becomes of great importance. This thesis researches the effect of the crisis on the importance of central bank communication. One would expect central bank communication to become more essential during the crisis because of the extraordinary measures taken by the ECB and other central banks to ease the effects of the crisis. The ECB has put great emphasis on communicating with the public and explaining their monetary policy decisions.

The research question is constructed as follow; Can you change the world by changing your words? The effect of ECB communication during a financial crisis. An analysis will be done on the ECB and the Euribor rates to answer the research question. Statements made by the ECB during their monthly press conferences, of a period of 11 years will be used and will later be split into two groups: the first being 6 years before the crisis, thus from 2004 till 2009 and the second group will contain data from 5 years during the crisis, thus from 2009 until 2014.

The remaining paper is structured as follows; the first section describes a review of past literature and the importance of communication. Then the data and method used in this thesis are explained in the second section. The third section consists of the regression results, and an explanation of the results obtained, the thesis ends with a conclusion.

4

Section 1: Literature review

1.1 The importance of transparency and communication

“Bernanke Speaks, and Shares Tumble”, reads the headline of the reaction of the stock market in the New York Times. Market participants analyze and react to the Federal Reserve’s words for signs of possible directions of monetary policy because it affects asset prices, more precisely, it affects stock prices (Kurov, 2012, 175-187).

Prior to the 90’s, central banks are believed to be shrouded in mystery. It was common knowledge that monetary policymakers should say as little as possible and say it in a secretive way (Blinder, Ehrmann, Fratzscher, De Haan, Jansen, 2008). For example, before 1994 the Federal Reserve never announced its federal fund rate, market participants had to speculate the target rate by analyzing the type and size of the open-market operations which were afterwards conducted by Trading Desk in New York to implement the policy (Woodford, 2005). Alan Blinder suggested a view in his 1998 paper of what central bank communications should be, a view that had been uncommon at the time but one that had been lurking around in the

underbrush. According to him, greater openness might actually improve the efficiency of

monetary policy because expectations about future central bank policy provide an important link between short- and long term rates. He also states that by providing the markets with more information about the central banks view of fundamental factors guiding monetary policy, the central bank makes market reactions more predictable to itself. More predictable market reactions make managing the economy more effortless (Blinder, Ehrmann, Fratzscher, De Haan, Jansen, 2008). Since the 1990’s the conduct of monetary policy has also changed in the United Kingdom. Reeves and Sawicki (2007, 207–227), quote King’s paper (2000) where he argues that mystery and mystique have given way to transparency and openness.

Communication is an essential tool to enhancing transparency because it is used to explain monetary policies and it provides information on the central banks assessment of economic developments, risks and uncertainties (Woodford, 2005). In this context transparency can be defined as “the absence of asymmetric information between policymakers and the public” (Ehrmann and Fratzscher, 2007, 509-541). Because the most essential decisions made in an economy are forward-looking, central banks not only affect the economy by setting an interest but also through their influence on expectations of future interest rate policy. As a consequence, there is good reason for central banks to explain their monetary policy decisions to the public and not only provide an explicit framework for decision making (Woodford,

5

objectives (Reeves and Sawicki, 2007, 207–227). It also helps financial markets anticipate central bankers’ actions more precisely (Musard-Gies, 2006). Thus Musard-Gies (2006) concludes that the response of interest rates to the announcement of macroeconomic data depends on the degree of transparency in the conduct of monetary policy. When participants of financial markets fully understand the causes of changes in inflation rates, how the central bank evaluates these influences and the steps it will take to deal with them, interest rates should adjust quickly to the information obtained by new macroeconomic data. According to the efficient markets theory asset prices should always reflect all available information, thus if this holds, then interest rates should adjust instantly after the publication of data that modify market participants’ expectations respecting monetary policy (Musard-Gies, 2006).

Woodford (2005), states the importance of clarity for market participants regarding the conduct of monetary policy, by saying that markets can to a large extent “do the central bank’s work for it,” the actual changes in overnight rates required to achieve the desired changes in incentives can be much more modest when expected future rates move as well. Kohn and Sack (2003) argue for a better understanding of monetary policy effects in order to achieve the

benefits of transparency. In other words, clear communication enhances the effectiveness of monetary policy (Woodford, 2005). But does this effectiveness become more crucial during a crisis? Before the regression results are explained, the types of ECB communication are explained and a brief overview of passed literature is presented proving that communication does influence certain economic variables.

1.2 Types of communication

The European Central Bank has several ways of communicating with the public to clarify how it will derive its goals and interpret its mandate. The following are communication channels adopted by the ECB:

o Monthly press conference: after the first meeting of the Governing Council every month,

the ECB President presents his introductory statement – explaining the monetary policy decisions. This is followed by a Q&A session with media representatives. Transcripts of the press conference are published on the website only a few hours later.

o The ECB's Monthly Bulletin provides the general public and the financial markets with a

detailed and comprehensive economic analysis. The publication date is usually one week after the first meeting of the Governing Council every month. It contains all the information that the Governing Council possessed in its meeting. It also contains articles providing

6

insights into long-term developments, into general topics or into the analytical tools used by the Eurosystem within the monetary policy framework. Furthermore, it contains a statistical section.

o European Parliament: the President appears once a year before the European Parliament

plenary to present the ECB's annual report. He also appears four times a year before the Parliament's ‘Committee on Economic and Monetary Affairs' (ECON) to explain the ECB's policy decisions and to answer questions by Committee members. The meetings are open to the public. Transcripts of the President's hearings at the ECON are published on both the Parliament's and the ECB's website. Other members of the Executive Board also appear occasionally before the Committee.

o Speeches by Governing Council members are an important tool for explaining the ECB's

view to the public. Speeches by the members of the Executive Board are published on this website. Interviews with Governing Council members are also an important communication channel.

o Visitors: the ECB receives a large number of visits from members of the general public and

experts from various institutions. A special visitor service takes care of general public groups.

o Dialogue with academic world: the ECB organises expert conferences & seminars. The

ECB publishes research results in the ECB's Working Papers and Occasional Papers Series.

o Statistics: data collected by the central bank are published, once their reliability has been

ensured. With the help of national central banks (NCBs) the ECB collects money and banking and related statistics, balance of payments statistics and international investment position statistics, and compiles financial accounts covering the euro area.

Source: https://www.ecb.europa.eu/mopo/strategy/comm/html/index.en.html

Once a month at 13.45, the ECB publishes its monetary policy decision by holding a short press release stating the interest rate decision. Forty five minutes later at 14:30 the ECB holds a press conference where the President of the ECB explains this decision in more detail. At the beginning of the press conference the President of the ECB reads an introductory statement that provides a comprehensive summary of the assessment of economic and monetary

developments and explains the monetary policy decision taken by the Governing Council. This part of the press conference is usually over by about 14:45. After the introductory statement journalists may ask questions relating to the ECB’s policy decision which are answered by the

7

President, (C. Brand, D. Buncic, J. Turunen, 2010, 1266–1298). The introductory statements presented by the ECB president during the monthly press conferences are used during this research.

Compared to other central banks communication approaches, the monthly press conferences by the ECB are far less detailed than for example the Bank of England (BoE) (Blinder, Ehrmann, Fratzscher, De Haan and Jansen, 2008).

1.3 Empirical evidence

Much evidence exists proving that central bank communication affects economic variables. Thus what central bankers say does actually matter. Short-term interest rates are directly

affected by the central bank through their monetary policy instruments (Musard-Gies, 2006). But Musard-Gies (2006) acknowledges the importance of long-term interest rates, stating that they influence consumption and investment decisions to a large extent. For example, mortgage rates, corporate bond rates and prices of long-lived assets such as houses, are affected by long-term rates rather than short-term rates (Musard-Gies, 2006). Blinder et al (2008), also argue the competence of a central bank to affect the economy depends on its ability to influence market participants’ expectations about the future path of policy rates, and not only on their current level. The logic behind this argument is simple, few economic decisions hinge on the overnight policy rates, and according to the expectation theory of the term structure, interest rates on longer-term instruments should reflect the expected sequence of future overnight rates (Blinder et al, 2008).

The key instrument of monetary policy is setting the interest rate on overnight loans between banks, which affects the quantity of excess reserves held by banks (J.D. Hamilton and J.C. Wu, 2012, 3-46). As seen in Graph 1 below, Euribor rates have been at historically low levels and near their zero lower bound since mid-2012. Low interest rates should decrease the amount of excess reserves however according to Hamilton and Wu (2012) the quantity of reserves remained unchanged and thus lowering interest rates even further offers little promise of stimulating economic growth. In other words interest rates as the primary policy tool of central banks, is not a tool designed to deal with the current crisis. Slowing growth and rising prices cannot just be solved by setting low interest rates (Morgan, 2009, 581-608). If a central banks primary tool cannot stimulate the economy, untraditional monetary policy tools must be used to provide macroeconomic improvement. One way for central banks to work around the zero lower bound constraint is by promising monetary accommodation in the future once the zero bound

8

ceases to bind (E.T. Swanson and J.C. Williams, 2014). Blinder et al (2008) also states that in the extreme case when interest rates decrease towards their zero lower bound, central bank communication about expected future rates becomes imperative to monetary policy.

Consequently, ECB communication should become the essence of monetary policy during the crisis. According to a study conducted by Hayo, Kutan and Neuenkirch (2008), central bank talk indeed becomes more relevant during the financial crisis between August 2007 and September 2009.

Graph 1: Euribor % change over a decade

Source: www.spanishpropertyinsight.com

In a study by Rosa and Verga (2006, 175-217) the effect of ECB communication on the price discovery process in the Euribor futures market is examined, it shows that financial markets react immediately to the ECB announcements, the policy decision announcement and the information released by the ECB president’s monthly press conference. Rosa and Verga also find that market interest rates can be explained by unexpected announcements, that is, the difference between what the ECB actually announces and what it was expected to announce. They prove this to be statistically significant, suggesting that ECB words have usually matched their deeds in the past. If they did not keep their word, there would have been no trust between market participants and the ECB and thus no one would react to what they say. (Rosa and Verga, 2006, 175-217). Their last finding suggests that the Euribor future market is efficient,

9

futures prices incorporate the news stemming from the ECB actions and announcements very quickly, in less than five minutes and around one hour respectively (Rosa and Verga, 2006, 175-217).

Another paper written by Rosa and Verga (2007, 146–175) analyzes and codes the introductory statement of the ECB presidents’ monthly press conference in order to determine how effective ECB’s words are in moving financial markets and whether or not their words are matched by actions. Once more they verify empirically that financial markets not only

understand but also believe the signals sent by the European monetary authority. Therefore, concluding that the ECB is effective in its job of communication to the public, and is able to influence market expectations on the short-term interest rate path using just words (Rosa and Verga, 2007, 146–175). Another finding by Rosa and Verga (2007, 146–175) suggests that the ECB can reinforce the effects of monetary policy actions by adding statements in the same direction, thus providing complementary rather than substitutable information with respect to macroeconomic variables.

A study by Musard-Gies (2005) tests whether ECB communication steer market short- and long-term interest rates in the euro zone. She uses a principal component analysis of euro-zone interest rates and concludes that short- and long-term interest rates react significantly to the bias in statements, especially to changes in statements from one meeting to the next. Musard-Gies (2005), also studied the individual reaction of short- and long-term interest rates finding that the short-term rates react more sharply than the long-term rates. Although her findings were all significant Musard-Gies (2005) suggests that the impact of monetary policy communication has to be judged in the light of other news events, which can have a much larger effect on the market, such as international developments, domestic macroeconomic data releases etc. (Musard-Gies, 2005).

Research on the FED and the BoE have also been conducted to prove the efficiency of central bank communication. Kohn and Sack (2003), investigate the effect of statements

released by the Federal Open Market Committee (FOMC) on market interest rates. Even though they conclude that central bank statements shape market expectations, Kohn and Sack (2003) state that monetary policy actions are without a doubt the essential factor for achieving certain goals of a central bank, for example; price stability. Nevertheless FOMC statements are still important and in addition to monetary policy actions they can be viewed as a vital ingredient of the monetary policy process (Kohn and Sack, 2003). They also suggest that when it comes to the short run, monetary policy actions and statements can serve as substitutes (Kohn and Sack, 2003). Reeves and Sawicki (2007, 207–227), investigate the impact Bank of England

10

communication has on the United Kingdom’s financial market. Their results are less strong than Kohn and Sack’s (2003) for the FED, but do conclude the significance of central bank

communication with the greatest impact of monetary policy statements on short sterling futures. The BoE communication offers an important avenue for policymakers to justify their decisions and help understand the Bank’s policy. In this way it is complimentary to interest rate setting, as well as helping the Bank of England fulfil its commitments to transparency and accountability (Reeves and Sawicki, 2007, 207–227).

Another variable affected by central bank communication are stock prices. Several researchers have analyzed the effect of monetary policy decisions on stock prices, and reached a consensus that stock prices strongly respond to unexpected decisions. Rosa (2011, pp. 915– 934) finds that both the unexpected component of policy actions and the unexpected

component of communication have statistically significant and economically relevant effects on stock prices, with the surprise component of FED’s statements explaining around 90% of the variation in S&P 500. These stock price movements also depend on the state of the economy (Kurov, 2012, 175-187). During a recession stock prices tend to fall whereas in a period of economic expansion stocks tend to rise following the FED's indications of lower rates ahead (Kurov, 2012, 175-187).

Adding to the importance of central bank communication is its ability to influence exchange rates. Research by Rosa (2011, pp. 478-489) shows that the unanticipated component of central bank statements and monetary policy action can clarify the changes of exchange rates. Moreover, only focusing on the current federal funds rate target misses the most important component of monetary policy, especially in recent years when market

participants have often well anticipated target funds rate changes (Rosa, 2011, pp. 478-489).

1.4 Problems with communication

The idea that transparency is unambiguously good is not universal (Reeves and Sawicki, 2007, 207–227). Amato, Morris and Shin (2002) argue that even though central bank communication is effective in steering market expectations, it also has its downsides. Agents tend to overreact to information and thereby enhance damage done by any noise, which in turn may crowd out private sector forecasting and thinking, discouraging the private sector from analyzing economic issues and forming an independent view (Amato et al, 2002, 495-503). An example of a misunderstanding happened in October 2000 when then-ECB President Wim Duisenberg hinted to an interviewer that there would be no further central bank intervention to support the euro. This led to immediate deflation of the euro and criticism towards Duisenberg (Blinder et al,

11

2008). If monetary policy statements are successful in steering expectation, then asset prices should react accordingly and policy decisions become more predictable, but it is not obvious that a central bank is always better off when saying more.(Blinder et al, 2008).

Section 2: Dataset

The data used in this experiment are the Euribor rates, GDP growth, HICP inflation and the monthly press conferences held by the ECB. Later a short description of each variable is given. The time period of 2004-2014 is chosen to compare the importance of communication before the crisis, 2004-2009, and during the crisis, 2009-2014. In total there are 130 monthly press conferences to be coded; in August 2004 and 2005 no press conference was held. The monthly press conferences are found on the following website:

https://www.ecb.europa.eu/home/html/index.en.html and the Euribor rates can be found on:

http://www.euribor-rates.eu/

First the Communication index is explained in subsection 1, followed by a interpretation of the remaining data and a brief data description and ending with an explanation of the empirical model used.

2.1 Communication index

In order to be able to run a regression on the statements they must have a numerical value. Each monthly press conference will be assigned a number between -2 and +2 according to the tone of the statements. The wording indicator variable can either be very dovish 2), dovish (-1), neutral (0), hawkish (+1) or very hawkish (+2). With the value of -1 indicating an easing period, thus a cut in interest rates is very likely and a value of -2 means the ECB is increasingly inclined to cut rates. The opposite of these two values are +1 and +2 indicating a tightened future monetary policy (Rosa and Verga, 2007, 146–175).

Blinder et al. (2008), state that coding approach is important because it helps us understand whether communication succeeds or fails. But this does not mean the coding approach is flawless. First downside of the coding approach is that it is subjective and there may be misclassifications. One way to reduce the risk of subjective coding is to let several researchers independently code each statement, but it will never be fully eliminated (Blinder et al. 2008). Another disadvantage according to Blinder et al. (2008) is that the coding is done ex post and might not accurately reflect how financial markets understood the signals during the press conference. The latter could be affected by e.g., expectations about monetary policy at

12

the time of the statement, because technically financial markets should only react to unexpected news presented during the press conference (Blinder et al, 2008).

Even though the coding approach is not entirely perfect it is still used to code the press conferences in this paper. The coding of the statements will be done the same way as it was done by Rosa and Verga (2007, 146–175), Table 1 shows how the statements will be ranked according to the words used in each one of them, followed by some examples of how the coding was done. If more than one word from Table 1 is used during a press conference, the overall code for the specific press conference will correspond to the mean of the indices of each single statement made by the ECB rounded to its nearest integer.

13 TABLE 1: Communication Index

Source: Rosa and Verga (2007, 146–175).

Most important keywords

Code

Tone

Imperative that upward pressure to be contained — Risks [to price

stability] are upward (upside) — The risks to price stability are

confirmed (or: remain) — Vigilant (vigilance) [with regard to upside

risks to price stability]— Close monitored (or: continuous close

attention) [upside risks] — Several [upward] factors need to be

monitored carefully

2

Very

hawkish

Both confident and vigilant (or: good however vigilant) [upside risks]

— Upward pressure remains contained — A number of (or: some)

upside risks need to be carefully monitored — Alert to emerging of

upward risks — Vigilance with regard to the materialisation of

upside risks

1

Hawkish

Appropriate — Favorable — Compatible — Consistent — In line —

Balanced — Absence of significant (or: no strong) pressures either

upwards or downwards — The downside risks have disappeared —

0

Neutral

Favorable, but there are some [downside] risks — Appropriate but

remain downside risks — Downside risks are not vanished — Some

of the downward risks had materialised

-1

Dovish

Consistent, but carefully monitor all [downside] risks to economic

growth; Balanced but monitor closely all [downside] factors; Monitor

carefully all [downside] factors relevant to economic growth —

Downside risks are still relevant — Economic slowdown is still

cause for concern — [Strong] downside risks for economic activity

— Monitor closely the downside risks to economic growth

-2

Very

14 Examples:

November 4th 2004: “Given the continued strength of M3 growth over the past few years, there remains substantially

more liquidity in the euro area than is needed to finance non-inflationary growth. This could pose inflationary risks in the future if the excess liquidity is not progressively reduced as a result of reverse portfolio shifts”

Code: 1

May 4th 2005: “Overall, when looking ahead, we see no significant evidence of underlying domestic inflationary

pressures building up in the euro area so that inflation rates should develop in line with price stability.”

Code: 0

May 4th 2006: “Against this background, the Governing Council will exercise strong vigilance in order to ensure that

risks to price stability over the medium term do not materialise.”

Code: +2

June 4th 2009: “We confirmed our expectation that price developments over the policy-relevant horizon will remain dampened by the marked weakening of economic activity in the euro area and globally.”

Code: 0

February 4th 2010: “A cross-check of the outcome of the economic analysis with that of the monetary analysis confirms the assessment of low inflationary pressure over the medium term.”

Code: -1

July 4th 2013: “It thereby provides support to a recovery in economic activity later in the year and in 2014. Looking

ahead, our monetary policy stance will remain accommodative for as long as necessary.”

Code: -2

December 4th 2014: “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”

Code: -1

2.2 The remaining variables

The three remaining variables are the Euribor rates, as the dependent variable and two control variables, economic growth and inflation are added.

For the interest rates the Euribor rates will be used. Euribor stands for Euro Interbank Offered Rate, and it is based on the average interest rates at which European banks borrow or offer to lend unsecured funds from one another (Euribor). The funds are traded in the euro wholesale money market, or the interbank market, and published by the European Banking Federation. Due to the fact that the Euribor rates provide the basis for interest rate swaps, futures, saving accounts and mortgages they are considered to be the most important reference rates in the European money market and thus are chosen as the interest rates for this

15

experiment. There are eight different Euribor rates from November 2013 and prior to November 2013 there were 15 (Euribor). For this experiment the 1 week, 1 month and 12 month average Euribor rates per month will be used. These rates can be found on the following website:

http://www.euribor-rates.eu/

The industrial production (IP) measures changes in the volume of goods produced, even though it only makes up less than 20 percent of the economy it is often used to measure GDP growth. The service industry on the other hand contributes much more to the economy but has a stable growth pace regardless of the state of the economy. IP is highly sensitive to changes in interest rates and demand, so it closely relates to changes in the overall economy (Baumohl, 2012). As a result, there is a close relationship between changes in industrial production and GDP growth and thus IP is used to represent the performance of GDP. Then yearly GDP rates are used but first calculated to monthly rates for a robustness check, to examine the stability of the regression model. The monthly HICP of 28 EU countries represent the inflation rate in this model. The rates are plotted in Graph 2.

Graph 2: HICP inflation rates

-1.0% 0.0% 1.0% 2.0% 3.0% 4.0% 5.0% 20 04J an 20 04J ul 20 05J an 20 05J ul 20 06J an 20 06J ul 20 07J an 20 07J ul 20 08J an 20 08J ul 20 09J an 20 09J ul 20 10J an 20 10J ul 20 11J an 20 11J ul 20 12J an 20 12J ul 20 13J an 20 13J ul 20 14J an 20 14J ul

HICP

16 -1.50 -1.00 -0.50 0.00 0.50 1.00 1.50 2.00 2.50 -1.20% -1.00% -0.80% -0.60% -0.40% -0.20% 0.00% 0.20% 0.40% 0.60% 20 04 20 04 20 05 20 05 20 06 20 07 20 07 20 08 20 08 20 09 20 10 20 10 20 11 20 11 20 12 20 12 20 13 20 14 20 14 GDP Comm

2.3 Data description

As said before the data can be split in two sub-periods, in this section the difference in tones between the periods are compared. Between 2004 and 2009, before the sovereign debt crisis emerged the tone of ECB communication is relatively more hawkish, and even very hawkish from 2006 until mid-2007. When looking at the GDP growth rate for the EU during this period it makes sense for a hawkish approach by the ECB. GDP growth rates were above three percent, increasing the risk for inflation. And thus pointing to a tightened future monetary policy. From 2009 until mid-2012, the ECB statements are more neutral and hawkish. From July 2012 ECB statements are on average dovish, indicating an easing period, which is logical when analyzing the GDP growth in this period. GDP growth was very low and sometimes even negative, in order to stimulate the economy one would expect a decrease in interest rates, and thus a dovish tone from the ECB. In Graph 3 the Communication Index and quarterly GDP’s are plotted. As seen in the Graph during times of slow economic growth ECB statements have been relatively more neutral and dovish, and when economic growth was high, 2006-2007, ECB statements were mostly hawkish.

17

2.4 The regression: Ordinary least squares

In order for the regression to be the best linear unbiased estimator six assumptions must hold. These are known as the six least squares assumptions in multiple regression:

1. The conditional distribution of 𝜀𝜀𝑖𝑖 given 𝑋𝑋1𝑖𝑖,𝑋𝑋2𝑖𝑖, … , 𝑋𝑋3𝑖𝑖 has a mean of zero.

This means that on average over the population of 𝑌𝑌_𝑖𝑖 falls on the population regression line. Therefore for any value of the independent variables the expected value of the residual equals zero. This assumption is crucial for unbiased OLS estimators. 2. (𝑋𝑋_1𝑖𝑖,𝑋𝑋_2𝑖𝑖, … , 𝑋𝑋_{𝑘𝑘𝑖𝑖}, 𝑌𝑌_𝑖𝑖 ), 𝑖𝑖 = 1, … , 𝑛𝑛, are independently and identically distributed (i.i.d.)

This assumption says something about how the sample is drawn. If the observations are drawn by simple random sampling from a single large population then

(𝑋𝑋2𝑖𝑖, … , 𝑋𝑋𝑘𝑘𝑖𝑖, 𝑌𝑌𝑖𝑖 ), 𝑖𝑖 = 1, … , 𝑛𝑛 are i.i.d. 3. Large Outliers are unlikely

Observations with values far outside the usual range of the date are unlikely. 4. No perfect multicollinearity

If one of the regressors is a perfect linear function of the other regressors this assumption is violated.

5. 𝑉𝑉𝑉𝑉𝑉𝑉(𝜀𝜀|(𝑋𝑋_𝑖𝑖) = 𝜎𝜎2

The variance of the distribution does not depend on 𝑋𝑋, thus the errors are homoscedastic.

6. 𝐶𝐶𝐶𝐶𝐶𝐶�𝜀𝜀𝑖𝑖, 𝜀𝜀𝑗𝑗� = 0, i ≠j.

The error terms are uncorrelated. Source: Stocks and Watson (2011)

If assumptions one through six hold then the estimators are BLUE. The model that will be tested is the following:

𝑖𝑖_𝑡𝑡+1− 𝑖𝑖_𝑡𝑡= 𝛼𝛼 + 𝛽𝛽₁𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝛽𝛽₂𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐷𝐷_{𝑐𝑐𝑐𝑐𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐}+ 𝛽𝛽₃𝑋𝑋₁+ 𝛽𝛽₄𝑋𝑋₂+ 𝜀𝜀 (1)

Where, 𝑖𝑖_𝑡𝑡+1− 𝑖𝑖_𝑡𝑡 , equals the change in Euribor rates between time t plus one month and time t. And where ‘t’ is a specific month and year between 2004 and 2014. The Communication Index is represented by 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 in the regression model. It can take the values -1,-2,0,1 and 2. In other words, how does a hawkish press conference held today (𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 = 1), affect the change in Euribor interest rates between today, t, and one month from now, t+1. The dummy variable,

18

𝐷𝐷𝑐𝑐𝑐𝑐𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐, is added to indicate whether the press conferences were held during the crisis or not. If 𝐷𝐷𝑐𝑐𝑐𝑐𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐= 1 the monthly press conference was held during the crisis and 0 if otherwise. If 𝛽𝛽2 is statistically significant and negative it indicates that during the crisis interest rates are further decreased if 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 has a positive value, and further increased if 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 is negative. Meaning that the crisis increases the importance of communication because it boosts the change in interest rates even more, but only if 𝛽𝛽₂ is statistically significant. The regressors 𝑋𝑋₁and

𝑋𝑋2represent control variables. Control variables are added because the communication index is not the only variable that plays a role in setting Euribor rates. Other variables such as economic growth and inflation rates also play a crucial role. The latter variables might be correlated to the main explanatory variable thus violating the first assumption of OLS regression. The remaining variable is, 𝜀𝜀, and exhibits the residual. Information on both the economic growth and HICP inflation rates in Europe can be found on http://ec.europa.eu/eurostat/home in the database.

The regression is run 3 times, first time using the change in 12 month Euribor rates, second time when using 1 month Euribor rates and a third time for the 1 week Euribor rates.

19

Section 3: The empirical analysis

Stata automatically assumes a homoscedastic variance of the residuals, meaning there should be no pattern between the residuals and the predicted values, if the variance of the residuals is not constant then the variance is known to be heteroscedastic. So in order for the results to be meaningful the residuals should be homoscedastic. One test on heteroscedasticity is the White’s test, this test is done on model 1 using the 12 month Euribor rates. White’s test obtains a p-value of 0.0000, rejecting the null hypothesis of homoscedastic variance of the residuals and thus proving heteroscedasticity. When adding robust to your regression in Stata, it assumes heteroscedastic standard errors and thus improving the calculated p-values. The White’s test is also done for the 1 month and 1 week Euribor rates, all proving heteroscedasticity. The

following results are obtained using robust standard errors.

3.1 Results from using 12 month Euribor rates.

After running the regression of model 1, with 129 observations a R-squared of 0.2639 is found. square is a statistical measure of how close the data are to the fitted regression line. A R-squared of 26.39 percent means that 26.39 percent of the change in 12 month Euribor rates can be explained by the independent variables used in model 1. In Table 1 the results are

presented.

Table 1: change in 12 month Euribor rates, robust standard errors.

| Robust

euri12m | Coef. Std. Err. t P>|t| [95% Conf. Interval]

---+--- comm | .0007994 .0001613 4.96 0.000 .0004802 .0011186 crisis | -.0003925 .0001718 -2.28 0.024 -.0007326 -.0000525 hicp | -.043435 .0267154 -1.63 0.107 -.0963123 .0094423 industry | .0638732 .0201054 3.18 0.002 .024079 .1036674 _cons | .0001531 .0003826 0.40 0.690 -.0006043 .0009104

20

Communication index: the coefficient of the communication index equals 0.0008, and thus has

a positive value. At a significance level of five and one percent, the communication index is indeed significant (p=0.000) and influences the change in Euribor rates. A positive beta for the communication index indicates that for hawkish central bank tones, the 12 month Euribor rates tend to rise by 0.08 percent and for very hawkish statements this rise is doubled. An economic explanation would be that relatively hawkish statements indicate a tightened future monetary policy, so the ECB is more likely to reduce the money supply in the future and thus rising interest rates. The opposite holds for relatively dovish statements.

Industry: the coefficient representing the GDP growth (industry) equals 0.0684, also positive. It

is significant with a p-value of 0.000 meaning the GDP growth is also one of the factors influencing the dependent variable. A positive coefficient for GDP growth can be explained by assuming that higher economic growth increases risks for inflation. In order to control the risks to inflation the ECB will increase interest rates to slow economic growth.

HICP: the coefficient for the inflation rate equals -0.0434, and thus has a negative relationship with the dependent variable. With a p-value of 0.107 the coefficient is not statistically significant at the five percent level. This indicates that the HICP inflation does not contribute to the

explanation of changes in Euribor rates. Over the past ten years HICP rates have been rather volatile as seen in Graph 2, having even experienced a time of deflation. This could be one of the reasons why this variable is not significant when predicting changes in 12 month Euribor rates.

Crisis: the crisis represents the communication times crisis-dummy variable in model 1:

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 ∗ 𝐷𝐷𝑐𝑐𝑐𝑐𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐, its coefficient is -0.0004 with a p-value of 0.024 this variable is statistically at a five percent level, indicating that the crisis does influence the importance of communication. When 𝐷𝐷_{𝑐𝑐𝑐𝑐𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐} = 1, the change in 12 month Euribor rates decreases further with 0.04 percent. A negative value for crisis is logical because when interest rates are decreased the economy is stimulated, hence contributing to economic growth. One would expect communication to become more important during the crisis becomes of the emphasis the ECB put on their unconventional monetary policy tools, one of them being forward guidance. Another reason for the importance to increase is the fact that interest rates have been at historically low levels and near the zero lower bound, then according to Blinder et al (2008) central bank communication becomes the essence of monetary policy.

21

The results indeed confirm Blinder et al (2008) theory of communication becoming crucial during turbulent times. They are also in line with findings by Hayo et al (2008) who confirm the

increased importance of FED talk during the financial crisis.

3.2 Results from using 1 month Euribor rates.

As it was shown before, the crisis does influence the importance of ECB communication when using 12 month Euribor rates, but does this result change when 1 month Euribor rates are used instead? After running model 1 with 1 month Euribor rates as dependent variable and using robust standard errors a R-squared of .2187 is found. Meaning 21.87 percent of the change in 1 month Euribor rates can be explained by the factors used in model 1. Thus model 1 explains the difference in 12 month Euribor rates better than the difference in 1 month Euribor rates. Again all independent variables except for HICP are statistically significant as seen below in Table 2. The interpretation of each independent variable is the same as before but with different values for each coefficient. The communication coefficient for 1 month Euribor rates is higher than for 12 month rates, which indicate that ECB communication has a higher impact on shorter Euribor rates.

Table 2:

---

| Robust

euri1m | Coef. Std. Err. t P>|t| [95% Conf. Interval]

---+--- comm | .0008738 .00022 3.97 0.000 .0004382 .0013093 industry | .0600418 .023625 2.54 0.012 .0132813 .1068024 hicp | -.0441453 .0285839 -1.54 0.125 -.1007209 .0124302 crisis | -.0005103 .0002507 -2.04 0.044 -.0010065 -.0000141 _cons | .0001032 .0004236 0.24 0.808 -.0007352 .0009417 ---

22

3.3 Results from using 1 week Euribor rates.

The last Euribor rate to examined is the 1 week rate. After running model 1 with 1 week Euribor rates as dependent variable an adjusted R-squared of .1788 is found. Meaning 17.88 percent of the change in 1 week Euribor rates can be explained by the factors used in model 1. Thus model 1 explains the difference in 12 month Euribor rates better than the difference in 1 month and 1 week Euribor rates. The results are presented below in Table 3.

Table 3:

| Robust

euri1w | Coef. Std. Err. t P>|t| [95% Conf. Interval]

---+--- comm | .0007341 .0002042 3.60 0.000 .0003299 .0011383 industry | .0524095 .0266494 1.97 0.051 -.000337 .1051561 hicp | -.0447921 .0296616 -1.51 0.134 -.1035007 .0139165 crisis | -.0003757 .0002939 -1.28 0.204 -.0009575 .0002061 _cons | .000206 .000461 0.45 0.656 -.0007065 .0011185 ---

When predicting the changes in 1 week Euribor rates none of the variables except for the Communication index is significant at a five percent level when using robust standard errors.

3.4 OLS assumptions

As was mentioned before in order for OLS to be the best linear unbiased estimator for this model a few assumptions must hold. In this subsection some of these assumptions are tested for the 12 month Euribor rates. Assumption number five, a homoscedastic variance of the residuals was already tested at the beginning of this section, here follow some other assumption testing.

Another important requirement is the normality of the residuals to ensure that the p-values are valid, if the residuals are not normally distributed then the p-values will not be binding and consequently the model will not be accurate. The residuals are plotted below in Graph 4.

23

Graph 4:

The residuals seem to be approximately normally distributed with a mean of roughly zero. Another crucial assumption for multiple regression is no perfect multicollinearity which can be tested using the VIF command in Stata. A mean VIF of 1.32 is calculated, normally VIF values of above ten are to be examined further, thus a value of 1.32 points to no threats of perfect multicollinearity.

3.5 Robustness check

In order to check robustness of model 1 a robustness check is performed. First a new model is introduced with split data. Then some variables are replaced by others to test the stability of the results. If the coefficients are indeed robust and stable, model 1 could be interpreted as a valid model. 0 1 00 2 00 3 00 D en s it y -.01 -.005 0 .005 Residuals

24

3.5.1 Results with split data.

One test to investigate the stability of model one is to split the data in two. Two similar

regressions are run on the two sub-periods. The first period runs from 2004 until 2008 and the second from 2009 until 2014. The model tested is:

𝑖𝑖𝑡𝑡+1− 𝑖𝑖𝑡𝑡 = 𝛼𝛼 + 𝛽𝛽1𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝛽𝛽2𝑋𝑋1+ 𝛽𝛽3𝑋𝑋2+ 𝜀𝜀 (2)

Where 𝑋𝑋₁ and 𝑋𝑋₂ represent the control variables HICP and GDP growth respectively and Comm stands for the Communication index. If the crisis becomes more important during the crisis 𝛽𝛽1should be more significant and have a bigger impact on the second period. Both regressions are run using robust standard errors. When running the regression the change in 12 month Euribor rates is used. A R-squared of 29.20 percent is found for the period prior to the crisis with 58 observations. The remaining results for the first period are shown below.

Table 4: change in 12 month Euribor rates, first period.

| Robust

euri12m1 | Coef. Std. Err. t P>|t| [95% Conf. Interval]

---+---

comm | .0015142 .0004853 3.12 0.003 .0005411 .0024872

industry | .0785386 .0384484 2.04 0.046 .0014543 .1556229

hicp | -.0533484 .062708 -0.85 0.399 -.1790703 .0723735

_cons | -.0008117 .0013797 -0.59 0.559 -.003578 .0019545

As seen in table 4 the variable HICP has no statistically significant meaning, thus in the period before the crisis it does not help explain the difference in 12 month Euribor rates. Preceding the crisis, communication and GDP growth are important factors when explaining the difference in 12 month Euribor rates, now the regression is run for the second period during the crisis with 72 observations. A R-squared of 4.9 percent is found, which is more than 20 percent less than the R-squared of the first period. The three independent variables are not significant at the five percent level meaning that none help explain the change in 12 month Euribor rates during the crisis.

25 Table 5: change in 12 month Euribor rates, second period.

| Robust

euri12m2 | Coef. Std. Err. t P>|t| [95% Conf. Interval]

---+---

comm | .0003209 .0003566 0.90 0.371 -.0003908 .0010325

industry | -.0118671 .01039 -1.14 0.257 -.0326 .0088659

hicp | .0060883 .0159562 0.38 0.704 -.0257518 .0379284

_cons | -.0009217 .0003314 -2.78 0.007 -.001583 -.0002604

Model 2 proves that ECB communication is not a factor in explaining changes in 12 month Euribor rates during the sovereign debt crisis, so it definitely does not become more important. A reason for non-significant variables could be the limited number of observations. Adding more observations to the regression makes the sample size more representative of the population which limits the influence of outliers or extreme observations (The importance of quality size sample), and hence ensure for a more precise model. Another reason could be the decrease of trust in the ECB, which has reached historical low levels during the crisis (Gros and Roth, 2009). According to Gros and Roth (2009) the decrease in trust in Germany, France and Italy, the three largest European economies, have been especially severe. Which explains why even though the ECB has put emphasis on forward guidance, it does not help move interest rates. For market participants to react to what is said during the monthly press conferences they should trust that the ECB’s words match their actions. An insignificant variable for communication suggests that people do not take ECB communication seriously, and so no one will react to what they say. As trust has fallen during the sovereign debt crisis communication will definitely not become more important during the crisis.

26

3.5.2 Results: variables replaced.

Another robustness check is done by replacing production in industry with yearly GDP growth rates in model 1. Using robust standard errors a R-squared of 22.75 percent is found and the variable communication is significant at a five percent level (p=0.003), but the crisis variable becomes insignificant with a p-value of .407. Also the HICP coefficient becomes statistically significant at the five percent level. Thus changing the GDP variable shows unstable results for the variable crisis and HICP.

Another robustness check can be done by adding the change in tone used by the ECB. Thus calculating the difference between tones for one month and the previous. The variable ‘deltacomm’ is added instead of comm, deltacomm represents the change in tone. An R-squared of .2170 is found, the results are shown in Table 6. The outcome differ from the ones reported in Table 1, disproving a robust model. Here the crisis again does not matter as in model 2 and HICP inflation is again not statistically significant.

Table 6: change in 12 month Euribor rates, robust standard errors.

---

| Robust

euri12m | Coef. Std. Err. t P>|t| [95% Conf. Interval]

---+--- deltacomm | .0009004 .0003708 2.43 0.017 .0001664 .0016344 industry | .0762766 .0214421 3.56 0.001 .0338368 .1187165 hicp | -.0019175 .0225416 -0.09 0.932 -.0465337 .0426986 crisis | .0000307 .000161 0.19 0.849 -.000288 .0003493 _cons | -.0001063 .0003575 -0.30 0.767 -.0008139 .0006014

---27

Conclusion

This thesis was written to research whether or not ECB talk becomes more crucial during the crisis. The empirical analysis was done using a multiple regression. The model used a dummy variable to represent the crisis period. Afterwards a second model was presented for a

robustness check, using split data from two sub-periods, the first being before the crisis and the second during. The empirical analysis shows three major conclusions.

First results show a significant level of the crisis and communication variable at the five percent level proving that ECB communication does influence the change in Euribor rates and an increasing importance of ECB talk during the crisis when using heteroscedastic standard errors, for both 12 and 1 month Euribor rates.

The second finding was a negative sign for the crisis variable. This indicates in times of a crisis, 𝐷𝐷_{𝑐𝑐𝑐𝑐𝑖𝑖𝑐𝑐𝑖𝑖𝑐𝑐} = 1, interest rates are further decreased which is logical because decreasing interest rates stimulate the economy and thus contributing to stronger economic growth.

Third, as said before the crisis-dummy variable is statistically significant for 12 month and 1 month Euribor rates but it is not significant for the 1 week rates. Model 1 explains the change in 12 month rates better than it does for the 1 month Euribor rates but the crisis variable itself has a greater influence on the 1 month Euribor rates. Using production in industry and HICP as control variables model 1 explains roughly 26 percent of the change in 12 month rates using robust standard errors.

These results coincide with the one found in Hayo, Kutan and Neuenkirchs paper who conclude that the financial crisis in the United States increased the influence of the Federal Reserve communication. On the contrary when the data is split in two, using robust standard errors, model two shows that ECB communication does not become more important during turbulent times, resulting in all variables being insignificant. Robustness check show unstable results. Findings change when variables are replaced with one another, concluding for an uncertain model.

Like Rosa and Vergas paper (2007) the ECB statements were coded according to their tone, which was either very hawkish, hawkish, neutral, dovish or very dovish.

With this research it is proven that ECB communication becomes more crucial during the sovereign debt crisis. Explaining the difference in 12 and 1 month Euribor rates the crisis does have an impact on the importance of communication. Thus central bankers should be careful about what they say in order to avoid miscommunication, especially during turbulent times. The question also remains how much information the ECB should signal to the public, since too

28 much transparency is not always beneficial.

The empirical analysis has a few shortcomings; first of all quantifying each statement by coding them is very subjective. Another drawback is the fact that this thesis only focused on the introductory statement of the monthly press conferences whereas there are a lot more ways the ECB communicates with the public which model 1 does not capture. In order to get an even better answer to the research question one should take into account all types of communication and then analyze their affect. A model for testing the importance of central bank communication during the crisis would ideally focus only on the surprise element of communication and thus control for other news affecting economic variables.

In further research it would be interesting to examine the impact of each different type of communication channel on the crisis, and add more variables that influence the change in interest rates. One could also research the importance of communication during a crisis on other macroeconomic variables that react to central bank communication.

29

References

1. Communication and monetary policy by Jeffery D. Amato, Stephen Morris and Hyun Song shin, 2002. Volume 8, No. 4, 495-503. Oxford Review of Economic Policy. 2. The EMU sovereign-debt crisis: Fundamentals, expectations and contagion by M.G.

Arghyrou, and A. Kontonikas. Volume 22, issue 4, 658-677. Journal of financial markets, institutions and money. October 2012

3. B. Baumohl (2012) The secrets of Economic Indicators: Hidden Clues to Future Economic Trends and Investment Opportunities. Upper Saddle River, New Jersey: Pearson Education, Inc.

4. Federal Reserve Policies to Ease Credit and Their Implications for the Fed's Balance Sheet by B.S. Bernanke at the National Press club Luncheon, National Press club, Washington D.C. February 2009.

5. Central bank communication and monetary policy: A survey of the theory and evidence by Alan S. Blinder, Michael Ehrmann, Marcel Fratzscher, Jakob De Haan and David-Jan Jansen, April 2008. Working paper 13932. National bureau of economic research.

6. The impact of ECB monetary policy decisions and communication on the yield curve by Claus Brand, Daniel BuncicandJarkko Turunen, 2010. Volume 8, Issue 6, 1266– 1298. Journal of the European Economic Association

7. Unconventional monetary policy in theory and practice by M. Cecioni, G. Ferrero and A. Secchi, September 2011. Bank of Italy Occasional Paper No. 102.

8. Communication by Central bank committee members: Different Strategies, same effectiveness? By Michael Ehrmann and Marcel Fratzscher. Volume 39, Issue 2-3, 509-51. March/April 2007. Journal of Money, Credit and Banking.

9. Forward guidance at the zero lower bound by Andrew Filardo and Boris Hofmann, March 2014, 37-49. BIS Quarterly Review.

10. The financial crisis and citizens’ trust in the European central bank. by Daniel Gros and Felix Roth. CEPS Working document No. 334, July 2010.

11. Do Actions Speak Louder Than Words? The Response of Asset Prices to Monetary Policy Actions and Statements by R S Gurkaynak, B Sack and E T Swanson, 2005. FEDS Working Paper No. 2004-66. International Journal of Central Banking

12. The effectiveness of Alternative Monetary Policy Tools in a Zero Lower Bound Environment by J.D. Hamilton and J.C. Wu, February 2012. Volume 44, Issue s1, 3-46. Journal of Money, Credit and Banking.

13. Financial market reaction to Federal Reserve Communications: Does the Crisis make a difference? By Bernd Hayo, Ali M. Kutan and Matthias Neuenkirch, January 2012. Joint Discussion Paper Series in Economics No 08-2008 University of Trier. 14. Is a word to the wise indeed enough? ECB statements and the predictability of

interest rate decisions. By David-Jan Jansen and Jakob de Haan, 2005. DNB Working papers No. 75.

15. Were Verbal Efforts to Support the Euro Effective? A High-Frequency Analysis of ECB Statements by David-Jan Jansen and Jakob De Haan, 2007. Volume 23, Issue 1, 245-259. European Journal of Political Economy

30

16. Central Bank Talk: Does It Matter and Why? By Donald L. Kohn and Brian P. Sack. FEDS Working Paper No. 2003-55 Board of Governors of the Federal Reserve. 17. What determines the stock market's reaction to monetary policy statements? By Alexander Kurov. Volume 21, issue 4, 175-187. Review of Financial Economics. 18. The European Sovereign Debt crisis by P.R. Lane, 2012. Volume 26, NO 3,

49-67(19). The Journal of Economic perspectives.

19. The limits of central bank policy: economic crisis and the challenge of effective solutions by Jamie Morgan. Volume 33, 581-608. Cambridge Journal of Economics. 20. Do ECB’s statements steer short-term and long-term interest rates in the euro zone?

By M. Musard-Gies, 2005. The Manchester School

21. Unconventional Monetary Policy of the ECB during the Financial crisis: An

assessment and new evidence by Christiaan Pattipeilohy, Jan Willem Van den End, Mostafa Tabbae, Jon Frost and Jakob De Haan, 2013. DNB Working Paper No. 381. 22. Do financial markets react to Bank of England communication? By Rachel Reeves

and Michael Sawicki, 2007. Volume 23, Issue 1, 207–227. European journal of Political Economy.

23. The high-frequency response of exchange rates to monetary policy actions and statements by Carlos Rosa, 2011. Volume 35, Issue 2, 478–489 Journal of banking & finance.

24. Words that shake traders: The stock market’s reaction to central bank

communication in real time by Carlo Rosa, December 2011. Volume 18, Issue 5, 915-934. Journal of Empirical Finance

25. The Impact of Central Bank Announcements on Asset Prices in Real Time: Testing the Efficiency of the Euribor Futures Market by C. Rosa and G. Verga, 2006. Volume 4, issue 2, 175-217 International Journal of Central banking.

26. On the consistency and effectiveness of central bank communication: Evidence from the ECB by Carlos Rosa and Giovanni Verga, 2007. Volume 23, Issue 1, 146–175. European Journal of Political Economy.

27. Stock, J.H. and Watson, M.M. (2011) Introduction to econometrics. International edition: Pearson.

28. Measuring the effect of the zero lower bound on medium- and longer-term interest rates by E.T. Swanson and J.C. Williams, September 2014. NBER Working Paper 20486

29. Central Bank Communication and policy effectiveness by Michael Woodford, December 2005. NBER Working paper No. 11898

30. The importance of quality sample size. Retrieved from:

http://www.uniteforsight.org/global-health-university/importance-of-quality-sample-size