The last two ECB meetings have resulted in the largest bond sell-off (June 5) and the largest bond rally (July 3) ever occurred on the day of a regular Governing Council meeting.
● This has brought the issue of communication back to the fore, ending the two-year long honeymoon between the ECB `and the markets and bringing back visions of the bad old times when a disconnect between the central bank and market players seemed to be the rule.
● The analysis developed in this paper casts some interesting light on central bank communication and on its nature, purpose, and pitfalls. The main conclusion of the analysis in our view are as follows:
● Central bank communication is an ongoing effort to gradually educate market participants on how the bank processes and reacts to macroeconomic data: communication gradually reveals the core of the policy reaction function. For the ECB we have shown that this is indeed the case.
● Data, and not words will be the main market movers in normal times. However, communication is crucial when there is a change in the policy reaction function. This in our view is exactly what happened in June. The abrupt market reaction indicates that the change in regime had not been communicated effectively, and points in our view to a weakness in the ECB’s communication framework. In case of a regime switch or a sudden change in the assessment of the economic outlook, intermeeting communication can play a precious role.
● Our analysis shows that ECB communication has been very effective in securing long-term predictability, whereas short-term predictability has broken down in recent months. This, however, might have partly been seen as the price to pay to enhance long-term predictability and antiinflation credibility. In other words, the June surprise might have been not just a communication dysfunction, but rather a conscious decision to shock the markets, to impress upon investors the seriousness of the ECB’s anti-inflation resolve.
Introduction – Back to the “bad old times”?
It all started with the June meeting. The unexpected and sudden announcement of a rate hike to be delivered the following month – unveiled in the Q&A and without resorting to the usual code words – and Trichet’s visibly hesitant posture brought the issue of communication back to the fore, ending the two-year long honeymoon between the ECB and the markets and bringing back visions of the bad old times when a disconnect between the central bank and market players seemed to be the rule.
Trichet’s ECB had been widely praised as a benchmark of transparency and efficient communication. By being predictable without pre-committing, through a wise and measured use of code words, the bank has helped markets steer interest rate expectations relatively smoothly. Even in recent turbulent times, the ECB was never considered a volatility generator. Until last June, only in 3 occasions a mild misalignment between ECB intentions and market expectations occurred. The first in November 2005 when two weeks before the December meeting Trichet had to flag that the coming rate hike was not priced in by markets. Then, in April 2006 Trichet said that the markets were wrong in pricing in a May hike. Finally, a few months later in July the President signaled the intention to speed up the pace of tightening by using the keyword vigilance one month earlier than expected. However, the market reaction to these episodes was negligible compared to what happened on June 5.
In terms of transparency, the ECB seemed to compare favorably to the Fed. With hindsight, it now seems that circumstances had as much to do with this as communication policy. The Fed was praised as the most transparent during its normalization phase: rates were clearly too low, they had to be increased significantly and gradually. Being predictable was an easy task. As the peak of the US hiking cycle approached and policy became more data dependent, full transparency became harder for the Fed to achieve just as the ECB instead went through its own normalization phase, where the path of policy was clearly traced, and the ECB became the new benchmark. However, the Fed was able to complete its tightening cycle, while the ECB’s was interrupted by the onset of credit crisis, which arguably made it even more difficult to switch gears.
On balance, the ECB’s communication effort since the onset of the crisis has been clear and effective—for example by clearly outlining the distinction between liquidity support measures geared to safeguard the functioning of financial markets, and monetary policy geared to macroeconomic developments, chiefly inflation. However, right at the beginning of the crisis we had the first warning sign that not all was well: in the run up to the September monetary policy meeting, a press release on monetary operations ended with the statement that the monetary policy stance had not been reassessed since the previous meeting. Since the previous meeting had used the code words “strong vigilance”, a number of market participants thought the ECB still intended to hike rates notwithstanding the onset of the crisis.
Unsurprisingly, the rate hike delivered last month against the background of a deteriorating growth environment attracted criticism and triggered heightened political pressure on the ECB. France’s Sarkozy proposed a more formal consultation between the ECB and the European Commission as well as the publication of the minutes of the Governing Council meeting. ECB Council member Bini-Smaghi replied in an article on the Italian financial daily Il Sole 24 Ore defending the ECB’s communication framework and Professors Francesco Giavazzi and Charles Wyplosz weighed in the debate saying that the ECB was right in hiking rates in July but that its communication strategy may definitely be improved from current levels.
There are two important issues at stake: the accuntability of the central bank, and the market impact of its communications. In this piece we concentrate on the latter, and try to assess the effectiveness of the ECB’s communication. We construct a simple communication index, and show that over the longer term the ECB’s communication has been highly effective in signaling the course of policy—deeds follow words in a rather consistent manner. Moreover, our communication index supports our view that around last June the ECB departed from its usual policy reaction function. We believe the shift could have been more effectively signaled, and argue that the lack of an effective inter-meeting communication is a pitfall of the current set-up. However, we also recognize that in June the ECB might have seen itself as facing a genuine trade-off between short-term and long-term predictability, and decided that surprising the markets was needed to bolster its anti-inflation credibility.
The communication challenge is far from over, however. We are still in a situation where macro indicators are volatile and at times inconsistent: while growth indicators are now decisively pointing south, it will take longer for inflation to come down to more comfortable levels, not to mention the risk of further volatility in commodity prices. Uncertain times require faster and more flexible policy reactions, and this might in our view require a more frequent guidance of market expectations than allowed by the regular statements. The ECB’s communication policy gets high marks so far, but the real challenge has just begun, and the risks, in terms of credibility and of market volatility, are high.
The first step of our analysis is the construction of an index to summarize and quantify the ECB’s assessment of the outlook for economic activity, prices and money in the eurozone. There are three different issues to consider in converting the ECB’s words into numbers: ● The type of communication, i.e. whether to include statements of all eurozone central bankers, only those from the GC members or just the press conference by the ECB’s President;
● What metric to use to convert words into numbers;
● How to deal with the use of “code-words”
As for the first point, we decided to consider only the introductory statement of the monthly press conference—although the recent literature also uses statements from all GC members and some studies include national central bankers1–for two main reasons. First, according to the ECB itself2, the introductory statement at the monthly press conference is the “principal vehicle of the ECB’s communication”. Plus, as the ECB follows a collegial communication approach, the content of these statements should reflect the views of all members of the GC in a way that supersedes statements by individual members. Second, statements by individual members could sometimes be directed to national audiences, and could therefore prove confusing for the markets (note that a key argument recently made against the publication of the minutes is that individual members could come under political pressure if their views were made public, which seems to suggest that attributable statements by individual members might not fully reflect their stance in the council—although they might provide some early insights on how views within the Council are evolving and how they will be expressed in future statements3).
Second, translating the words in the introductory statement into a quantitative measure implies a trade-off between the need to have a simple and replicable rule and a truthful and correct interpretation of the statement. While an approach based on mechanical rules or simple word-counting has the advantage of being easily replicable, it disregards the so-called “reading between the lines”. As we believe central bank watching is an art as well as a science, relying too much on a mechanical algorithm might come at the high cost of losing information or even misinterpreting central bankers’ words (the same word or expression can have different meanings, according to the economic juncture). Our approach, detailed below, tries to allow for a sufficient degree of subjectivity while at the same time extracting a reliable and clear signal.
Third, over the recent tightening cycle, ECB President Trichet made large use of “codewords”: “strong vigilance” would signal a hike at the following meeting, when it would be replaced by “monitor closely”, with “vigilance” indicating a new hike was on its way but not imminent. While these code words were useful in preparing the markets for changes in the refi rate and in helping the ECB to become more predicable and accountable, they were mainly a feature of the last tightening cycle, so that the evidence is too short to incorporate them in a quantitative measure of the central bank stance extracted from their communication strategy.
Weighing words: Constructing an index of communication
A central bank’s rhetoric is by itself qualitative in nature, so that any attempt to translate words into numbers is difficult and to some extent inevitably arbitrary. This holds true particularly in the eurozone context, as the ECB has to deal with a number of different media from different countries and different backgrounds. The most intuitive and direct criterion for a measure is whether the informed reader perceives the wording of the statement as being dovish or hawkish. We therefore follow the methodology developed in Berger, de Haan and Sturm (2006)4 and categorize the message on a scale from –2 (extremely dovish, hinting at lower rates in the near future) to +2 (extremely hawkish, pointing to higher rates). We apply this grading to the three sections of the statement, i.e. economic activity, prices and money/liquidity. The first two constitute the economic analysis (or first pillar), whereas the last represents the monetary pillar, which is a distinctive feature of the ECB’s communication and strategy.
To limit the degree of subjectivity, several of us read and graded the ECB statements separately and independently. In order to even out the subjective differences and to get a single signal, we then employed principal components analysis (PCA) and extracted factors from each of the three groups (economic activity, inflation, money). As the first factor of each group tends to explain around 90% of the total variance, we can safely use it as the sub-index for inflation, economic activity, and money.
Finally, in order to get a unique measure of communication, we applied again PCA on the three sub-indices: the first factor explains 87% of the total variance and hence can be adopted as our index of ECB communication.
We do realize that such a measure is fraught with weaknesses: it is still highly subjective, and most importantly it is constructed with the benefit of hindsight, that is as we read the old statements now we know what policy actions followed, which might influence our reading today. Having said that, we still believe the index is informative and allows us to get some interesting conclusions, as we argue below.
Let deeds follows words!
A first graphical inspection reveals that our communication index tends to lead ECB’s actions on the refi rate, something that is also confirmed by a series of Granger causality tests. A simple regression shows that the communication index leads the refi rate by 3-6 months, with the coefficient being positive and highly significant, meaning the future monetary policy actions are an increasing function of today’s words. As we said above, this might partly be because the index is constructed reading the statements with the benefit of hindsight, but it does seem to confirm that the ECB’s rhetoric is a very transparent indicator of policy direction.
Which message do markets listen to? If the ECB’s rhetoric as conveyed by the statements is such a powerful predictor of rate moves, then should we expect every statement to be market moving? We already know from experience that the answer is no, but let’s first check it analytically.
Following the approach developed in the event study literature5, we regress the change in yields ( Δr ) on the change in the ECB Communication index ( ΔIndex ) according to the following equation:
Δr =α + βΔIndex +ε
We focus here on money market and short-term rates, as they are the yield curve buckets most directly influenced by monetary policy. We thus use the 3M Euribor and the 3M Eonia rates, plus the 2Y yield on the German Schatz. For each maturity, we took the difference in yield between the closing value of the day before the meeting and the closing value of the day in which the meeting was actually held. ΔIndex is the difference between Index at time t and Index at time t-1, thus expressing how ECB’s rhetoric has changed from one meeting to the following one.
The above equation suggests that if markets are surprised by the change in the tone of ECB communication, β should be positive and significant. As a matter of fact, all our sets of regressions show that this is not the case, as β turns out to be not significantly different from zero for all interest rates we have used.
This of course is exactly as it should be, and is in fact the sign of a successful communication policy. Central banks normally try to steer market expectations gently and gradually (although there are cases where shock therapy is called for). In an ideal situation, the regular flow of central bank communication over the longer term should be a transparent predictor of policy moves and steer market rates in the right direction, but each individual statement should only add moderate incremental information. The communiquè should help the market gain a proper understanding of the monetary policy framework the central bank is pursuing: as Blinder (2002)6 pointed out, “perhaps the best a central bank can do is to teach the market its way of thinking”. Thus, through the statement the ECB gives the market the tools to read the current juncture “through its own eyes”: once the market is equipped with such tools, the risk to misinterpret the correct working of the reaction function should be minimized.
A smooth communication strategy will be of course easier to achieve in normal times than in times of crisis, where policy may need to adapt quickly to rapidly changing circumstances. The result is fully consistent with the bulk of the literature showing that the “surprise” element contained in the announced policy decisions has progressively become smaller. ECB scholars7 show that out of a total of 144 days, markets were “surprised” only eight times, including the emergency measures adopted in the aftermath of 9/11. Then June 2008 came…we will discuss this later.
Using communication to improve the policy signal
Before we go any further, we want to conduct another test of the communication’s informational content and predictive value. In order to do that, we resort to our proprietary Taylor rule based on PMI-output gap and M3 growth8. The first step is to see how communication can improve our understanding of the policy reaction function. We thus estimated four different equations, adding to our original specification the level of our communication index and the sub-indices for activity, inflation, and money. The results shown in the table below suggest that the overall communication index helps explain movements in the policy rate, as the coefficient is positive and highly significant.
While in all four cases the overall fit of our model improves only marginally, the analysis of the residuals reveals that including the sub-index for inflation (rather than the overall index) delivers the best performance in terms of forecasting error over the last six months. This is the most interesting result of the analysis: it suggests that the very tough antiinflation rhetoric of recent months signaled a deviation from the ECB’s usual policy reaction function. We will return to this point in the conclusions. However, our analysis also warns that rhetoric alone cannot act as a substitute for actual data: the fifth column in the table above shows that a Taylor Rule based solely on our communication sub-indices in substitution for actual data on growth, inflation and money, delivers worse results than the ones based on data. This would suggest that while words are important, as they allow a better understanding of future ECB’s actions, actual figures should not be dismissed so easily. To put it differently: ECB communication helps market participants to understand the bank’s policy reaction function, but it is then crucial for any ECB watcher to monitor the data closely and feed them correctly into the reaction function.
Concluding remarks – A return to recent consistency is crucial
We believe the analysis developed in this paper casts some interesting light on central bank communication and on its nature, purpose, and pitfalls. We have developed a quantitative indicator of ECB communication that is a good leading indicator of policy rate moves, that improves the predictive power of our Taylor rule in the most recent period—albeit without substituting for the actual macro data—but at the same time does not predict market rate moves. The main conclusion of the analysis in our view are as follows:
Central bank communication is an ongoing effort to gradually educate market participants on how the bank processes and reacts to macroeconomic data: communication gradually reveals the core of the policy reaction function. Over a longer period, effective communication should therefore correlate Highly with actual policy: deeds should follow words. For the ECB we have shown that this is indeed the case.
Data, and not words will be the main market movers in normal times. Once market participants have a good sense of the central bank’s reaction function, they can guess how the bank is likely to react to incoming data. Central bank communication is still extremely important as a regular check of how the bank is reading the data. As central bankers tend to smooth the data rather than overreacting to individual data points, communication should reveal only gradual changes in the bank’s assessment of the economic outlook. We should not forget, however, that this assessment can at times differ significantly from that of the markets (the Fed for example has at times been engaged in a clear tug of war with the markets).
Communication is crucial when there is a change in the policy reaction function. This in our view is exactly what happened in June. The ECB’s heightened emphasis on inflation developments signaled a departure from the usual reaction function, which our analysis suggests relies heavily on the output gap and which would not have suggested a hike. The commodity price shock and the attendant risk of second round effects changed the rules of the game, and effective communication became essential.
The fact that the last two press conferences have triggered the sharpest bond sell-off (June 5) and the strongest bond rally (July 3) ever occurred on a Governing Council’s meeting day indicates that the change in regime had not been communicated effectively, and points in our view to a weakness in the ECB’s communication framework. In case of a regime switch or a sudden change in the assessment of the economic outlook, inter-meeting communication can play a precious role. In the case of the ECB, however, we have seen that communication between meetings is often a confusing cacophony, which leaves market participant unable to distinguish those council members who are voicing personal opinions for the benefit of national audiences from those who might be signaling a shift in the consensus view of the Council. This is not to say that council members should always speak with one voice; to the contrary we believe that airing different views is a good way to allow some insight in the discussions held within the council. But this should still leave room for some key council members to signal ongoing shifts in the thinking of the consensus. Our impression is that the ECB is gradually moving in that direction, but it will still take some time to finetune the strategy.
A further issue is the trade-off between short-term and long-term predictability, highlighted in an excellent recent ECB analysis9 cited above. The first refers to the markets’ ability to correctly anticipate near-future moves on the part of the central bank; whereas the latter indicates the “public’s general understanding of the overall process of monetary policy making, resulting in the ability to anticipate the broader future direction of monetary policy”. As we argued above, our analysis shows that ECB communication has been very effective in securing long-term predictability, whereas short-term predictability has broken down in recent months. This, however, might have partly been seen as the price to pay to enhance long-term predictability and anti-inflation credibility. In other words, the June surprise might have been not just a communication dysfunction, but rather a conscious decision to shock the markets, to impress upon investors the seriousness of the ECB’s anti-inflation resolve. Everybody out there – including barking politicians – now knows that unmoored inflation expectations and lax wage deals won’t pass unnoticed in Frankfurt. The price, however, has been a significant bout of volatility in already unsettled financial markets.
The communication challenge is far from over, however. We are still in a situation where macro indicators are volatile and at times inconsistent: while growth indicators are now decisively pointing south, it will take longer for inflation to come down to more comfortable levels, not to mention the risk of further volatility in commodity prices. Uncertain times require faster and more flexible policy reactions, and this might in our view require a more frequent guidance of market expectations than allowed by the regular statements. The ECB’s communication policy gets high marks so far, but the real challenge has just begun, and the risks, in terms of credibility and of market volatility, are high. Especially if oil prices’ recent trend becomes more entrenched, then GDP softness will gain the upper hand over stilluncomfortable inflation prints. The ECB will progressively soften its current sanguine growth view but what is more important for the central bank is to make sure that the last two meetings remain isolated within a consistent communication framework. Undesired volatility is the last thing markets need these days.
1 See for example, F. Heinemann and K. Ullrich (2005) “Does it pay to watch central bankers’ lips? The information content of ECB wording”, ZEW working paper; D.J. Jansen, J. DeHaan (2006) “Does ECB communication help in predicting its interest rate decisions”, CESifo working paper 1804.
2 See ECB (2007). “Communicating monetary policy to financial markets”, ECB Monthly Bulletin, April 2007.
3 See Jansen and De Haan (2005) “Were Verbal Efforts to Support the Euro Effective? A High-Frequency Analysis of ECB Statements”” DNB working paper 033. They show that statements by individual officials on interest rates, inflation and economic growth have been often contradictory, thus reducing their power as leading indicators.
4 See H. Berger, J. De Haan, J.Sturm (2006) Does money matter in the ECB strategy? New evidence based on ECB communication, CESifo working paper No. 1652
5 Krueger, J. T. and K. N. Kuttner (1996), “The Fed Funds Futures Rates as a predictor of Federal Reserve policy”, Journal of Futures Markets, 16(8)
6 Blinder, A. (2002), “Through the looking glass: central bank transparency”, CEPS Working Paper No. 86.
7 See T. Blattner, M. Catenaro, M. Ehrmann, R. Strauch, J. Turunen (2008). “The predictability of monetary policy” ECB Occasional Paper n°83, March 2008
8 See “A PMI based Taylor Rule for the euro area” A. Maccario, M. Valli, UniCredit Research Series N°8, November 2006
9 See T. Blattner et al. (2008), op. cit.