«CENTRAL BANKS’ TRANSPARENCY: WORDS AS SIGNAL FRANCESCO CENDRON GIANFRANCO TUSSET University of Padova April 2014 “MARCO FANNO” WORKING PAPER ...»
UNIVERSITÀ DEGLI STUDI DI PADOVA
Dipartimento di Scienze Economiche ed Aziendali “Marco Fanno”
CENTRAL BANKS’ TRANSPARENCY:
WORDS AS SIGNAL
University of Padova
April 2014 “MARCO FANNO” WORKING PAPER N.178 Central banks’ transparency: words as signals Francesco Cendron and Gianfranco Tusset*
Jel Classification: B59, E58 Keywords: central banks; transparency; language; speeches; content analysis.
Introduction Ross Starr, a general economics and financial scholar, wrote that “money, like written language, is one of the fundamental discoveries of civilization” (Starr 2012: 1). This paper takes money and language together and focuses on the language of central bankers in order to gain better understanding of its defining effect on monetary policy making and the overall communication strategy which it backs up.
This paper examines words – particularly words uttered by central bankers in their speeches – in order to grasp how central bankers treat, consider, and discuss crucial features of their policies. We will investigate how relevant words are used in these speeches and how they can be ‘grouped’ together to create semantic clusters. Among these words, we will focus on ‘transparency’ in
particular, since it is a crucial concept marking the recent evolution of central banks’ practices:
Transparency represents the most dramatic difference between central banking today and central banking in earlier periods (Dincer and Eichengreen, 2008: 105) We will analyze how ‘transparency’ is presented and used by central bankers and not filtered by economic theory. Our paper focuses on how central banks perceive transparency and how transparency is expressed in the speeches of central bankers. We believe that words can reveal a central bank’s marginal changes as an institution that adapts itself to an evolving economic and financial world. Words can map a central bank’s evolution.
As said, transparency will be our ‘key word’, since it is indicative of a concrete transformation that has occurred, at least at communication level, within central banks. To bring the topic into * Francesco Cendron is an independent researcher: firstname.lastname@example.org; Gianfranco Tusset is at the University of Padua: email@example.com. The authors are grateful to participants at the Palermo meeting and the Livorno conference for their useful comments, to Marisa Cemin, Francesco Purificato, Arjuna Tuzzi, and to two anonymous referees for their suggestions. Finally, our especial thanks go to Noam Chomsky for his helpful and encouraging words.
The authors you are grateful to MIUR for its financial support: 2009NXTCP9 focus, it is useful to begin with the most accepted academic view of this concept. From a theoretical
perspective, Petra Geraats defined transparency as follows:
Central bank transparency could be defined as the absence of asymmetric information between monetary policy makers and other economic agents. This means that it reduces uncertainty and this is often believed to be beneficial … Furthermore, transparency may affect the incentives that policy makers face to manipulate private sector beliefs through signaling and reputation building (Geraats 2002: F533).
Consequently, at first sight: “The aim of central bank transparency is to lessen or eliminate informational asymmetries between central bank decision-makers and the private sector” (De Haan et al. 2007: 4). But, given a context of asymmetric information, transparency in the procedures used to assess the private sector’s expectations improves output stabilization (Eijffinger 2004), which may be one of the objectives of the monetary policy. So what is the aim of transparency?
It must also be established whether the reduction of information asymmetries is the main aim of transparency. The notion itself of an optimal level of transparency, which is now widely debated (Geraats, 2002; Van der Cruijseen, Eijffinger, & Hoogduin, 2010), is meaningful inasmuch as it refers to a target: “It seems to be optimal to have an intermediate degree of transparency by limiting some forms of transparency” (Van der Cruijseen, Eijffinger, & Hoogduin, 2010: 1483). If the literature agrees that increasing transparency improves economic performance, the concept of an
intermediary level arises for different reasons, among which the following:
The first reason is that a lot of transparency could lead to uncertainty. By providing too much information, people start to focus too much on the complexity of monetary policy making and the uncertainty surrounding forecasts. While the actual quality of their forecasts might not be affected, agents perceive the quality of their forecasts to be worse. The second reason is that a high degree of transparency could lead to an information overload and confusion. The assumption that individuals are capable to absorb, understand and weigh all the information that the central bank provides is probably too strong (Van der Cruijseen, Eijffinger, & Hoogduin, 2010: 1483).
According to this quote, we can assume that central bankers themselves question the optimal level of transparency – once they agree on what transparency refers to, of course. Most of the recent economic literature on central bank transparency is devoted to demonstrating that more transparency or a given level of transparency reduces inflation bias and expectations of inflation in general: “There is an optimal degree of central bank transparency at which inflation persistence is minimized” (Van der Cruijseen, Eijffinger, & Hoogduin, 2010: 1484). Although we do not dispute the importance of the transparency/inflation relationship, we shall consider transparency from a wider perspective that is open to different relationships and meanings, and which takes speeches into account as part of a more general communication strategy of the central bank.
Owing to its superior informational endowment, and according to the rational expectations paradigm, central bank’s communication is a linear process between the sender and the receiver.
When rational expectations are formed cooperatively between the central bank’s communications and their receivers, the central bank’s communication strategy corresponds to its monetary policy in the long run (Drexter, 2012).
Our analysis is carried out before considering expectation–making, and it gives insight into the sentiment of the message conveyed by the speeches. Transparency becomes speech sentiment, which will become expectations, and will feed back linearly to the speeches themselves. We are not interested in the reception and feedback process of the communication strategy: we are interested in how transparency is treated in the speeches in terms of language and not as a communication tool.
Speeches are seen as objects of the central bank’s rhetorical discourse: they can comment on any stage of the monetary policy decision making process, which corresponds to a specific key feature of central bank transparency (Geraats 2002). The paper also considers the potential receivers of the transparent monetary policy communication, i.e. the financial and real economy. Speeches are not a stand-alone tool of monetary policy: time-inconsistency opportunities and money-illusion are relevant only as a counter-reaction of all economic agents to the central bank’s communication strategy. Since the overall monetary policy is considered symmetric with the overall communication strategy, we break the speeches down into relative frequencies in order to understand trends in the central bank as a communication sender and focus specifically on its language.
Regardless of the optimal degree of transparency and its effect on inflation, we focus on the rhetorical style of the central bank as a communication sender, not forgetting that speeches may be processed, managed and fed back by the receivers.
Section 1 introduces the method, the corpus, and the interpretation criteria. Section 2 shows how central bank independence and transparency are considered in the speeches, and how this somehow mismatches with the current transparency indexes. Section 3 focuses on inflation targeting regime and its connection with transparency. Section 4 demonstrates how the financial economy is considered in the central bankers’ speeches, while Section 5 goes into details on the real economy and speeches. The paper ends with some concluding remarks.
1. Method This paper is based on data obtained with tools not commonly used in economics and history of economic thought. A principal component analysis and a correspondence analysis are the main techniques employed for multidimensional factorial analysis. The nature of these tools is exploratory and not probabilistic. The analyses yield a data matrix which enables interpretation of the relationships between row profiles or column profiles built by assuming given sets of variables (central banks or years). This view is contextual in the sense that it displays relationships as they exist simultaneously over a broad corpus of texts by reducing the amount of information in order to show the data in a variety of structural directions.
The Bank for International Settlements (BIS) collects and publishes the speeches of senior central bankers on its website, which is from where the final corpus was retrieved. It includes all of the 7837 speeches of central bankers from the 39 institutions1 considered in this study during the period from 1997 to 2012. The corpus was analyzed according to the variables ‘year’ and ‘central bank.’ Basic statistical measurements show that this was a very large corpus. Words occurring with a low frequency (less than 50) were removed, but particular interesting segments2 were included to investigate specific economic areas. Examples include financial stability, inflation targeting and real economy.
The analysis of this big corpus, or the ‘big data,’ required the use of specific software. We used Taltac to manage the corpus and Spad to extract the figures.
Figure 1 and Figure 2 are the two basic graphs on which we based our study.
Australia, Botswana, Brazil, Bundesbank, Canada, Chile, China, Denmark, ECB, England, Finland, France, Hong Kong, India, Indonesia, Italy, Japan, Kenya, Korea, Malaysia, Mauritius, Mexico, Netherlands, New Zealand, Nigeria, Norway, Pakistan, Philippines, Saudi Arabia, Singapore, South Africa, Spain, Sweden, Switzerland, Thailand, Turkey, Uganda, United Arab Emirates, FED.
A ‘segment’ is composed of several words expressing a single concept. Examples: ‘rate of interest’ or ’financial risk’.
Figure 1. Years 1997-2012 (Inertia: Axis 1: 28,84%, Axis 2: 15,09%)
Figure 1 illustrates the distribution of the words and segments (the ‘black dots’) used by central bankers from 1997 to 2012 and which are considered in this study. In a sense, this figure shows the ‘evolution’ of the central bankers’ lexicon. It is clear that words and discourses change.
Figure 2. 39 central banks (Inertia: Axis 1: 8,78%, Axis 2: 6,98%)
Figure 2 shows words and segments grouped around the 39 central banks treated here. The software did not allow us to consider a number of central banks above this amount. The final graphical visualization represents the initial corpus in terms of statistical associability between elements. Thus, it provides an advanced text analysis tool with which to derive high-quality information from complex corpora. Since we investigate 39 central banks, we obtain 38 axes that display the words investigated according to the banks considered, and the axes can be combined to form different plans. Each plan represents part of the variability, and it shows a single synthetic outcome based on the part of variability that it represents. The small amount of inertia (roughly speaking the variability of words over the set of 39 banks) represented by the two most important axes in Figure 2 indicates that the variability is distributed quite well over the various banks: this means that each bank carries and plots a different rhetorical style.
What do these graphs represent? The interpretation criteria are well known in the relevant
literature (Tuzzi, 2003; Bolasco, 2013):
• the position of words in the graph is meaningful only if considered in the global context generated by the full corpus of all words used by all banks or over the full period;
• if two words are close, this means they are similar in terms of being used with the same relative frequency by all banks or over the full period;
• if two banks are close, this means that they are similar in terms of using the same vocabulary with the same relative frequency;
• words and banks (as well as words and years) can be studied in terms of statistical associability according to the angle that they form with the axes. The more similar the angle is, the more statistically associated the elements are;
• the further the words are from the origin of the graph, the more they have determined the context which the graph represents.
Finally, this paper also includes a quantitative analysis of the simple relative frequencies for words appearing in the whole corpus.
2. Independence, transparency and speeches