DO EUROPEAN CENTRAL BANK’S STATEMENTS STEER INTEREST RATES IN THE EURO ZONE?* potx

DO EUROPEAN CENTRAL BANK’S STATEMENTS STEERINTEREST RATES IN THE EURO ZONE?* by MARIE MUSARD-GIES IXIS Corporate Investment Bank and University of Orleans LEO In this study, we aim at te

DO EUROPEAN CENTRAL BANK’S STATEMENTS STEER

INTEREST RATES IN THE EURO ZONE?*

by

MARIE MUSARD-GIES IXIS Corporate Investment Bank and University of Orleans (LEO)

In this study, we aim at testing whether press conferences held after themeeting of the European Central Bank’s monetary policy council steermarket interest rates in the Euro zone To meet this goal, we quantify thestatements according to whether they are neutral, hawkish or dovish Weshow, using a principal components analysis, that market interest ratesreact significantly to the bias in statements, and more particularly tochanges in statements from one meeting to the next Moreover, we findthat the short end of the yield curve reacts more sharply to statementsthan the long segment: the effect of statements peaks on interest rateswith a maturity of 6 or 12 months and is smaller for the longer maturities.Using non-parametric tests confirms our previous results

it had done: the decision to leave interest rates unchanged was perfectlyexpected by the financial markets, but the FOMC’s decision to delete thesentence ‘policy accommodation can be maintained for a considerableperiod’ and replace it by ‘the Committee believes it can be patient in removingits policy accommodation’ was interpreted by the financial markets as asignal that the Fed was going to tighten its monetary policy faster than whathad been previously anticipated

In recent years, the general move in central banks to enhance theirtransparency has had the consequence of improving substantially the predict-ability of monetary policy decisions Several papers document the extent towhich US monetary policy has become increasingly open and transparentand how these moves towards greater openness and transparency haveincreased the ability of markets to anticipate policy actions Thus, Poole and

* I would like to thank Patrick Artus and Florence Beranger for their many valuable comments and suggestions.

1 Statements reported by Bernanke (2004b).

© 2006 The Author

Journal compilation © 2006 Blackwell Publishing Ltd and The University of Manchester

Published by Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK, and 350 Main Street, Malden, MA 02148, USA.

116

Rasche (2000) and Poole et al (2002) investigate the extent to which market

participants anticipate Federal Reserve policy actions Their most importantfinding is that not only is the market better able to anticipate funds rate targetchanges, but it appears that the market is able to anticipate such changes

further in advance In more recent papers, Lange et al (2003) and Swanson

(2004) conclude that a higher degree of transparency of the Fed is connectedwith a higher degree of predictability In the Euro area, Perez-Quiros andSicilia (2002) find that market interest rates have predicted Euro area interestrates comparatively well up to three months in advance.2

Transparency of monetary policy allows financial markets to betteranticipate the monetary policy decisions As a result, the response of interestrates to the publication of macroeconomic data depends on the degree oftransparency in the conduct of monetary policy The theory of efficientmarkets predicts that the prices of financial instruments will always reflect allavailable information If markets are efficient, interest rates should adjustvirtually instantaneously after the release of data that modify financialmarkets’ expectations concerning monetary policy Transparency thereforecauses financial markets to adjust their interest rate expectations as soon asmacroeconomic data are published, in advance of any action by the centralbank In this vein, Haldane and Read (2000) show that a reduction in themarkets’ uncertainty about the central bank’s reaction function implies thatmarket prices will react less to monetary policy changes since market partici-pants are better able to anticipate them and more fully to news about the state

of the economy, in particular macroeconomic data releases on which thereaction function is conditioned Consequently, markets react to macroeco-nomic announcements they view as important arguments to the monetarypolicy reaction function and, moreover, react more strongly to those unan-ticipated data releases that have greater impact on potential future monetarypolicy Thus, in a world where the central bank’s reaction function wasknown to the market participants with certainty, one would in principleobserve no financial asset price reactions at the time of monetary policychanges, but significant reactions to the release of surprise macroeconomicdata that occur before the monetary policy action date

Insofar as monetary policy decisions are now largely predictable, andconsequently well expected, one should ask what the role of central banks is inthe implementation of monetary policy if financial markets are themselves able

to digest and factor the new information into interest rates Do central bankshave the possibility to make monetary policy more effective? Clear communi-cation helps to increase the predictability of monetary policy decisions, andthus causes financial markets to adjust their interest rate very quickly and well

2 According to their approach, over the period between 4 January 1999 and 6 June 2002, which included 78 meetings of the Governing Council of the European Central Bank, the market correctly anticipated 94 per cent of the decisions.

© 2006 The Author

before the meeting on monetary policy Nevertheless, are central banks able to

go further in moving asset prices in the desired direction?

Communication of the central bank has a real importance in this tion (nearly perfect predictability) because it enables monetary policy to bemore efficient Indeed, to the extent that communication provides usefulguidance to markets about the future path of short-term interest rates, centralbanks will exert greater influence over the longer-term interest rates that mostmatter for spending decisions Actually, setting of short-term interest rates bythe central bank has no more than a small impact on the investment decisions

situa-of the private sector inssitua-ofar as the latter depend primarily on the level situa-oflong-term interest rates However, the link between expectations about mon-etary policy and long-term interest rates is well known, e.g theories about theterm structure of interest rates such as the theory of expectations Conse-quently, expectations about monetary policy are at least as important as thecurrent level of short-term interest rates in terms of determining long-terminterest rates—hence the role played by the central bank’s communicationpolicy because the way it communicates is how it will be able to give itsinterpretation about trends in economic activity In other words, if it iscredible, the more the central bank provides information to the marketsabout how it assesses trends in inflation, in real activity etc., the more theexpectations of financial markets and of the central bank will tend to alignthemselves with one another and, ultimately, the more the central bank willinfluence long-term interest rates

In this paper, we study the effect of European Central Bank (ECB)communication on interest rates of different maturities More precisely, weaim at testing whether the statement made during the press conference thatfollows the announcement of the ECB’s decision about the main refinancingrate, for its part, has an impact on interest rates To do so, we are going tolook whether the tone of the ECB’s statement (which we are going to codify)

or the change in the tone from the previous statement explains changes in theEuro zone’s short- and long-term interest rates We briefly review, in Section

2, empirical studies on central bank communication Section 3 then discussesthe issue of how to measure communication This is followed by our empiri-cal analysis of the effectiveness of ECB statements in influencing Euro zoneinterest rates in the desired way in Section 4 Section 5 presents the results.Section 6 concludes

2 Empirical Studies on Central Bank Communication

The empirical literature3 on central bank communication is quite small,partly reflecting the difficulty of measuring it, partly due to the relatively

3 References related to theoretical models estimating the impact of communication are older and numerous We can refer to surveys on the transparency: the most recent is Carpenter (2004); other surveys are those of Geraats (2002) and Hahn (2002).

© 2006 The Author

recent adoption of transparency as a major characteristic of central bankpolicy These papers analyze the effect of central bank communication onasset prices Most studies test whether communication affects exchange rates:Jansen and de Haan (2003) find some effect from ECB statements on thevolatility of the Euro and Fratzscher (2004) finds more systematicevidence in favor of effectiveness for the three G3 monetary authorities inchanging the level and volatility in the desired direction.

We will focus here on the impact of ECB communication on the yieldcurve The first paper to analyze the effect of Fed communication on marketrates is Kohn and Sack (2003) These authors use daily data and show thatwhen the Fed holds a speech (statements that can be three types of commu-nication: statements by the FOMC Chairman Greenspan on the day of theFOMC meeting, testimonies and other speeches of Greenspan), then marketrate variance (which corresponds to the volatility of the error term in regres-sions) is much stronger This suggests that financial markets react to state-ments delivered by the Fed Furthermore, Kohn and Sack (2003) distinguishtwo types of statements, depending on their contents: one referring to the

‘monetary policy inclination’ and the second one to the ‘economic outlook’.These authors conclude that statements by Greenspan about the monetarypolicy inclination have a significant effect on the volatility of short-terminterest rates while statements about the economic outlook tend to have a

significant impact on longer maturities In the same vein, Bernanke et al (2004) and Gürkaynak et al (2004) find that US financial markets attribute

considerable importance to statements that include an indication about thefuture path of policy

Ehrmann and Fratzscher (2005) analyze the communication strategiesand assess their effectiveness for three central banks: the Fed, the Bank ofEngland and the ECB They focus on forward-looking policy statements(speeches, interviews and testimonies) delivered by all policy-makers (notonly the central bank’s governor) distinguishing communication on meetingdays from inter-meeting statements Following the terminology also used byKohn and Sack (2003), these authors decided to keep the categorization assimple as possible They conclude that US markets react significantly morestrongly to statements by Greenspan and less to statements by other FOMCmembers, whereas Euro area markets respond to communication by the ECBPresident and other Governing Council members to a very similar extent.Finally, they find that US markets react to statements both about monetarypolicy inclination and the economic outlook, whereas UK and Euro areamarkets respond mostly only to communication about monetary policy.4

Finally, the paper of Rosa and Verga (2005), in a similar vein to ourstudy, analyzes the communication content of ECB press conferences These

4 This difference probably reflects, according to Ehrmann and Fratzscher, the different market perceptions of policy reaction functions.

© 2006 The Author

authors construct an indicator to capture inflation and real economy risksand conclude that market expectations at different maturities (from one to sixmonths) react to ECB communications This paper covers a short-term timehorizon (from October 2001 to September 2004) that captures only a period

of increasing interest rates Moreover, the authors have limited their analysis

to the meetings in which the ECB did not change the main refinancing rate.All these studies conclude that central bank communication has signifi-cant influences on the expectations of financial markets The study of Rosaand Verga (2005) is most closely related to our approach We extend thisliterature by analyzing the effect of ECB statements on the yield curve inorder to assess if ECB communication affects the short end and the long end

of the yield curve differently

Contents of ECB’s Statements?

In this section, we turn to the issue of how to measure communication As ourobjective is to test whether and to what extent ECB’s statements affect marketinterest rates, on the day of the press conference that follows the announce-ment of the ECB’s decision about the main refinancing rate, we need toconstruct an indicator that quantifies the content of ECB communication.Contrary to Kohn and Sack (2003) and Ehrmann and Fratzscher (2005)who distinguish between ‘monetary policy inclination’ statements and ‘eco-nomic outlook’ statements, we have codified all the statements made at pressconferences from 1999 to October 2004 (i.e a statement per month generallyspeaking) by drawing a distinction between statements with a ‘hawkish’ (i.e.statements that seemed to indicate that future policies might involve higherrates than previously thought), ‘very hawkish’, ‘neutral’, ‘dovish’ (i.e state-ments that seemed to indicate that future policies might involve lower rates

than previously thought) or ‘very dovish’ tone An indicator variable (DECB)takes the values+2, +1, 0, -1 and -2 according to the tone of the statement

At each press conference, the ECB discusses the prospects with respect tohow prices will trend in the medium term (as its main objective is medium-term price stability) via several dimensions: it analyzes and directly antici-pates trends in consumer prices (moves in energy prices, prices of food goods,wages etc.) but also in real activity and in the money supply via growth inmonetary aggregate M3.5 In its introductory statement, the ECB thereforepresents its inflation and growth scenarios, as well as the implicit (upside ordownside) risks for its central scenarios It is by drawing on these scenarios

5 A noteworthy point is that the structure of the press conference changed from May 2003 onwards From 1999 to April 2003, risks weighing on medium-term price stability were analyzed by drawing on two pillars (pillar one, trends in M3; and pillar two, a collection of indicators having an impact on prices) Subsequently, from May 2003 onwards, the two pillars were replaced by economic analysis and monetary analysis This does not modify our codifying work, however.

© 2006 The Author

and associated risks that we ascribe a ‘rating’ to the statement (e.g a scenario

of growth equal to its potential with upside risks and a rise in inflation andwith also upside risks in the medium term will be deemed very hawkish).Rosa and Verga (2005) use specific expressions or code words that areused frequently in the ECB press conference in order to translate the quali-tative information into an index Nevertheless, their indicator relies only onthe ‘synthetic judgments’ part of the introductory statement6and most of thetime the synthetic judgment is only a small part of the whole communication.Consequently, this quantification by only some expressions has some draw-backs: some very important but uncommon expressions can be missed by anautomatic analysis Moreover, a precise coding would require an analysis ofthe grammatical structure of sentences: indeed, meaning and strength ofkeywords often depends on the context of the sentence

Another study is the one of Gerlach (2004): it relies on the editorial of theECB monthly bulletin This analysis is a more subjective one: he does notconstruct a glossary of words based on a few lines only to construct an index.For each editorial, using information synthesizing the overall reading of theeditorial, he allocates a different value to the three dimensions: inflation, realactivity and M3 Consequently, this approach allows reading ‘between thelines’ and therefore is more subjective than the systematic approach of Rosaand Verga (2005)

Our study is based on two coding principles merging these precedingideas First, one is like the Rosa and Verga (2005) study with some slightdifferences: it relies on the automatic analysis of the keywords of the wholeECB press conference (and not only on the synthetic judgment) A corpus ofvocabulary often found in ECB statements has been defined automatically(by analysis of the statements with a computer), and each word or expression

of this corpus has been assigned a weight (negative in the case of a dovish one,positive in the case of a hawkish one) The corpus is used to approximatelyquantify the tone of the communication However, precision of this codifi-cation is limited for the reasons detailed above Consequently, this firstcoding, which can be considered as a global trend of the tone of the ECBstatements, is refined in order to improve the quality of the analysis We use

an additional coding system more related to the Gerlach (2004) approach.The automatic coding defined before is improved when some important

‘triggering’ sentences are present (not only keywords, the whole sentences areimportant to understand that there is a real change in the tone) For example,the press conference of February 2000 is coded as+1 and the press conference

in March 2000 is coded as+2 because in March the ECB added the ing’ sentence: ‘the Governing Council concluded that vigilance is required’ (it

‘trigger-6 Rosa and Verga provide in their paper a description of how an ECB press conference is organized: after some greetings, there is a synthetic judgment on the risk for price stability, followed by a judgment on growth, inflation and monetary variables At the end of the introductory statement, the synthetic judgment is repeated.

© 2006 The Author

was the first time that the ECB used this word to express its concern aboutinflation risk) This explains why we changed the status of this press confer-

ence to very hawkish Actually, what we are coding in this second step is the

change in the tone from one meeting to the other one (‘differential’ coding).Obviously, this operation is more subjective and for this reason coding hasbeen done by two persons, comparing and merging the results after.Our study covers a time horizon from January 1999 to October 2004,7

which yields 66 observations (on the other hand, the Rosa and Verga (2005)study is based on only 30 observations) The final codification we obtain ispresented in Appendix A (Table A1) We compared (Appendix A, Fig A1)our codification of statements with that carried out by Gerlach (2004) Wetake the sum of the ratings set by Gerlach or calculate a weighted average(with larger weight for ‘activity’ and ‘inflation’ ratings, i.e 40 per cent, thanfor the rating relative to M3, i.e 20 per cent) We conclude that our assess-ment of ECB statements is quite similar to the one drawn upon by Gerlachwhen we look at the weighted average of his ratings The only major differ-ence concerns 2004, when ECB statements were relatively hawkish in ouropinion, while he deems them to have been neutral

Note that the tone of ECB statements (Appendix A, Table A2) is moreoften hawkish than accommodating even though, in four out of the six years

of observation, growth in the Euro zone was lower than its potential growthrate (for the ECB, potential growth is close to 2–2.25 per cent) Simulta-neously, the inflation target has exceeded 2 per cent every year except in 1999(and inflation is the objective of the ECB’s monetary policy)

4 Data and Methods

We aim at testing whether information from the ECB’s press conferenceshave effects on financial market expectations, i.e whether day-to-day change

in short- and long-term interest rates around the ECB meeting is related tothe tone of the statement We use approaches that are common in the ‘event-studies’ of finance literature: ordinary least squares regressions analysis,

where ECB statements are represented by our dummy DECBthat codifies thetone of the statement, and non-parametric tests in order to test the robustness

of our previous results

The studies conducted in the USA and reviewed in Section 2 use intradaydata and therefore assess the impact of the Fed’s statement in the minutes justafter the statement.8We use daily data since our objective is to test whetherthe statements have a durable impact on interest rates It is normal that themarkets should react to a macroeconomic figure or a statement: conse-

7 No press conference is held in August Furthermore, two press conferences were held in March and October 2000, and this explains why there were 13 press conferences in 2000 instead of

11 in the other years.

8 Only Kohn and Sack (2003) use daily data.

© 2006 The Author

quently, asset prices move in the wake of announcements However, what wewould like to ascertain is whether the initial reaction lasts a few hours and isalways factored into interest rates at the end of the day To do so, wecalculate, for each interest rate we consider, the difference between the inter-est rate on the day of the monetary policy council meeting and the interestrate on the day before the Governing Council meets (closing price) to testwhether ECB communication affects the term structure of interest rates Inother words, for each ECB meeting, we calculate the day-to-day reaction ofinterest rates at different maturities.

Moreover, we are interested in the effect of ECB communication on theEuro zone’s short- and long-term interest rates, in order to test whether theeffect of ECB communication is different, depending on the maturities Forour analysis, we have therefore chosen to focus on several money marketrates, in other words, the one-month Euribor, three-month Euribor, six-month Euribor and 12-month Euribor spot rates With respect to long-terminterest rates for the Euro zone, we use the price of German futures contracts(which are the benchmark of the Euro zone yield curve), two-year (Schatz),five-year (Bobl) and 10-year (Bund) rates A future contract is a bindingagreement between two parties to make a particular exchange on a specified

date t in the future.9

The interest of working on contracts (for the longsegment) rather than spot rates lies in the fact that, generally speaking,futures are far more reactive (and thus factor in any additional informationfar faster) All these data (Euribor spot rates for the short end and futurescontract for the long end of the yield curve) can be downloaded fromDatastream.10

In the event-study literature, authors regress the change in asset prices onthe change in policy rate:11

whereDR tstands for the change in asset prices andDk tstands for the change

in monetary policy rate For example, in Cook and Hahn (1989),Dk tstandsfor the change in the Fed Funds target rate: they examined the day-to-dayresponse in the USA of bond rates to changes in the target Fed Funds ratefrom 1974 through 1979 The response to target rate increases was positiveand significant at all maturities, but smaller at the long end of the yield curve.Results for more recent periods show a much weaker relationship between

9 A future contract is very similar to a forward contract, which is also a contract to trade on a future date The main differences are that futures are always traded on an exchange, whereas forwards always trade over the counter Furthermore, futures are highly standard- ized, whereas each forward is unique.

10 The future price is quoted on a daily basis for the delivery months March, June, September and December A future continuous series can be calculated from the traditional traded months:

we use the continuous series (calculated by Datastream) that smoothes the series during the switch over period from one month to another based on trading activity.

11 The sample consists only of days of central banks’ meetings.

© 2006 The Author

target rate changes and other interest rates (e.g Roley and Sellon (1995)apply the Cook and Hahn event-study approach to the 1987–95 period).Similarly weak results for the 1989–92 period were obtained by Radecki andReinhart (1994) This apparent deterioration of the relationship betweentarget rate changes and market interest rates can be explained by the generalmove in central banks to enhance their transparency This is why Kuttner(2001), in a context of enhanced transparency of central banks, has perfectedthe approach of Cook and Hahn Using the Fed Funds Futures to identifythe expected and the unexpected component of the monetary policy decision,

he documents a much stronger relationship between market rates and expected changes in the funds rate target

un-Finally, if we suppose that monetary policy is perfectly predictable, thesurprise on the day of the monetary policy meeting is no longer provided bythe decision about the policy rate, but rather by the content of the statement

of the central bank Indeed, in the Euro zone, as underlined in the tion of this paper, market interest rates have predicted Euro area interestrates comparatively well up to three months in advance.12

introduc-Bernoth and vonHagen (2004) conclude that the policy decisions of the ECB have been pre-dictable on average: they show that, since May 2001, markets were notsurprised by the decisions on the rates of the ECB

However, we have to note that in May 2001 the ECB decision waslargely unexpected As soon as early 2001, the markets were expecting arate cut by the ECB Nevertheless, the ECB did not change its key interestrate in February, March, or even in April 2001, whereas the economic slow-down seemed to justify a rate cut (inflation was admittedly still high despitethe fall in oil prices and was picking up again in March–April but this wasmainly the result of the mad cow disease, i.e an external supply shock).Even as the markets were banking on a rate cut, the ECB’s statementsremained neutral The fact that its statements did not change from onemonth to the next should not have led to fluctuations in interest rates andyet they were trending downwards: at this point in time, the marketsbelieved in economic indicators more than in the ECB In fact, it eased itsmonetary policy in May, thus comforting the markets, while still makingrather neutral statements, as inflation had precisely peaked in this month atits highest level since the launch of the European Monetary Union at 3.1per cent (but 3.4 per cent according to its measure at the time, which wassubsequently revised)

Consequently, in order to take into consideration the fact that a fewmonetary policy decisions were not perfectly expected, we estimate the day-

12 In this regard, the decision of the Governing Council of the ECB in November 2001 to switch from bimonthly to monthly discussion of monetary policy may have affected the predict- ability of the ECB, as the timing of its interest changes can be anticipated more easily by the market.

© 2006 The Author

to-day change of interest rates as a function of two different components:first, the unexpected component of the monetary policy as introduced byKuttner (2001), and second, the informational content of central bank state-ments Rosa and Verga (2005) include in their regression the actual change

in the ECB policy rate on the day of the press conference For the reasonsexplained above, the actual change in the ECB main refinancing rate is notconsistent with the fact that markets are able to anticipate most of ECBdecisions before the meeting This is the reason why we prefer to include ameasure of the unexpected change in the policy rate (the ‘surprise’ in mon-etary policy decisions) rather the actual change in policy rate the day of theECB meeting This yields the following equation:13

where DECB stands for our dummy that quantifies the tone of ECB’s

state-ments and S stands for the unexpected component of the monetary policy

decision (called the ‘surprise’ of monetary policy)

The question, then, is how to extract a measure of the surprise Weneed to use forward interest rates in order to extract financial market expec-tations More precisely, we will look at forward interest rates one weekbefore the day of the press conference and compute a surprise as the dif-ference between the ECB main refinancing rate on the day of the meetingand the forward rate one week ahead A forward interest rate is an interestrate that is specified now for a loan that will occur at a specified futuredate.14

A standard assumption holds that a forward interest rate is the sum oftwo components: first, a liquidity premium (also called a term premium);second, an expectation concerning the spot rate that will hold at the time

Thus, a one-week rate one week forward of x per cent might be considered to

be a consensus expectation of market participants that the one-week spot rate

will equal x per cent in one week (the liquidity premium is considered to

be insignificant for a maturity of one week) The underlying concept of thisassumption is the so-called expectations hypothesis of the term structure: twoequivalent investment options should have the same expected return, other-wise investors would arbitrage away any differences With the exception of aterm premium, there should be no difference in the returns from holding along-term bond or rolling over a sequence of short-term bonds To compute

a forward rate, we use the following formula:

13Note that the index t does not have a specified frequency The subscript t denotes the day of a

meeting (and only these days).

14 The interest rate is termed a forward interest rate to emphasize the fact that it covers an interval that begins at a date forward (i.e in the future).

© 2006 The Author

where f is the forward rate (assuming a 360-day basis), d1is the number of

days from the settlement date to the start date of the forward period, d2thenumber of days from the settlement date to the end date of the forward periodand FV the future value The formula of FV is as follows:

+

1

3601360

2 2

1 1

r d

where r1stands for the spot rate for d1days and r2for the spot rate for d2days

It would be relevant to compute a one-week rate one week forward(having the same maturity as the main refinancing rate): we would need theone-week and two-week Euribor spot rates to compute this one-weekforward rate Unfortunately, the data for the two-week Euribor spot rate areavailable only from October 2001 Consequently, we will compute a three-week rate one week forward by using the one-week and the one-monthEuribor spot rates (by assuming that the difference between one-week andthree-week interest rates is insignificant) The measure of the surprise in

policy rate on the day i of the press conference is given by

where the index i denotes here the day of the ECB meeting and has a daily frequency and r istands for the new (i.e after the decision of the GoverningCouncil) ECB main refinancing rate on the day of the ECB press conference

Finally, f i-7is the three-week interest rate one week forward, one week beforethe ECB meeting Hence, equation (2) will be tested empirically with thecalculation of the surprise above

5 Results: Market Reactions to ECB Communication

We now turn to the question whether ECB’s statements influence financialmarkets by moving market interest rates in three phases We carry out aprincipal components analysis (PCA) in order to test first whether ECBcommunication affects financial markets and then whether the short end ofthe yield curve reacts more sharply to statements than the long segment Thisboils down to estimating equation (2) for the short end and the long end ofthe yield curve

Then, after analyzing whether communication has an effect on marketinterest rates, we will assess over what horizon financial markets are beingaffected: consequently, we will estimate separately the impact of ECB’sstatements on each market interest rate (i.e for all available maturities).Finally, we will use non-parametric tests to confirm the robustness of pre-vious results

© 2006 The Author

5.1 PCA on Short- and Long-term Market Interest Rates

We use a PCA on day-to-day changes in Euro zone interest rates Euribor, 3M-Euribor, 6M-Euribor, 12M-Euribor, Schatz, Bobl and Bund) toextract the common information contained in interest rates around the date

(1M-of the statement PCA consists in projecting the n daily changes in the interest

rates we consider (Euro zone short- and long-term interest rates) on the basis

of n vectors (orthogonal with one another) Centering and reducing data

prevents the more volatile series from ‘crushing’ the estimate Furthermore,this enables us to interpret the relative weight of each interest rate in the axesderived from our PCA Initially, we carry out a PCA on all interest rates(short- and long-term interest rates), and then we subsequently carry out aPCA on short-term interest rates exclusively and then on long-term interestrates

5.1.1 Interest Rates React Far More to the Change in the Tone from One Statement to the Next Than to the Statement in Absolute Terms. When wecarry out the PCA of changes in short- and long-term interest rates, we obtain

a first factor that explains 52 per cent of the variance of all the changes inshort- and long-term interest rates This factor is well linked to all the changes

in short- and long-term interest rates: the weights of each market interest rate

in the first principal component range between 0.34 and 0.43 if we look at thefirst eigenvector (Table 1) which means that the series are weighted in avirtually identical manner in this first factor Consequently, this first factorsatisfactorily represents the common moves in short- and long-term interestrates in the Euro zone

We now estimate via ordinary least squares the relationship between thefirst factor, called PC1, derived from our PCA and our variable DECB thatcodifies the statement between -2 and +2 (equation (6)) The estimationobtained is presented in Table 2.15

15Note that the value of the coefficient of our dummy DECB cannot be interpretable economically since we have used changes in short-term interest rates but also from the opposites of changes in prices of contracts on long-term interest rates, bearing in mind that all data are centered and reduced.

T able 1

E igenvectors from PCA