Printer Friendly

Does the clarity of Central Bank communication affect volatility in financial markets? Evidence from humphrey-hawkins testimonies.


This article investigates whether greater clarity of central bank communication coincides with lower levels of volatility in financial markets. Recent empirical work has established that central bank talk is often an important source of information for the private sector. Surveying the literature, Blinder et al. (2008) conclude that "communication can be an important and powerful part of the central bank's toolkit since it has the ability to move financial markets."

Still, it is not clear what would constitute an optimal communication strategy. In practice, large differences exist in the way in which central banks communicate. Some central banks, such as the Federal Reserve and the Bank of England, publish detailed minutes of their monetary policy meetings, whereas others, such as the European Central Bank, do not. At the same time, the European Central Bank is one of the few central banks holding an elaborate press conference shortly after the interest rate decision. Finally, some central banks, such as the Reserve Bank of New Zealand and Norges Bank, publish forecasts of their own policy rate, whereas most monetary authorities refrain from doing so.

Alongside the issue of optimal communication, there has been a long-standing debate on central bank transparency. A greater degree of transparency may have important benefits, such as well-anchored inflation expectations. At the same time, various authors have suggested that limits to transparency may exist (Cukierman and Meltzer 1986, Morris and Shin 2002). Whether transparency is beneficial, and to what extent, is in the end an empirical matter. The evidence in favor of transparency has accumulated over the years. Although Geraats (2002) cautiously concluded that the available evidence "suggests that transparency tends to be beneficial," more recently, van der Cruijsen and Eijffinger (2007) find that "The empirical results are largely in favor of more transparency."

Many studies in the literature use measures of institutional transparency, such as whether the central bank publishes macroeconomic forecasts or information on its policy deliberations. So far, the clarity of actual central bank communications has received little attention. (1) Still, if the central bank communicates regularly, but does so in a nonaccessible manner, it can hardly be called transparent. Moreover, the relationship between clarity and volatility seems intuitive: when central bankers communicate clearly, financial markets have less difficulty in interpreting the central bank's message. This leads to less uncertainty, which is reflected in lower levels of volatility.

Therefore, this article contributes to the literature by focusing on the effects of clarity of communication. For a broad range of financial instruments, it is tested whether clarity of communication affects volatility. Two readability statistics are used--the Flesch reading ease score and the Flesch-Kincaid grade level--to measure the clarity of central bank communication. Both statistics have been widely applied in other fields. (2) One benefit of these two criteria is their objectivity: they are completely based on the characteristics of the underlying texts. Readability statistics are applied to the testimony by the Chairman of the Federal Reserve at Congressional Monetary Policy Oversight hearings--more commonly known as the Humphrey-Hawkins hearings. For almost 30 years these testimonies have been one of the main communication channels of the Federal Reserve.

This article has three key results. First, when clarity matters, the results point to the conclusion that clarity diminishes volatility. Second, clarity of communication matters mostly for volatility of medium-term interest rates. Third, the effects of clarity vary over time. There is an indication that clarity has especially mattered during Alan Greenspan's term at the Federal Reserve. Overall, the analysis shows the importance of transparent communication on monetary policy.

This article proceeds as follows. Section II discusses the relevant literature on transparency and communication. Section III discusses the methodology and the data, while Section IV gives the results. Sections V and VI consider whether the effect of clarity has varied over time. Section VII offers my conclusions.


Over the last two decades central banks have undertaken important steps toward more transparent communication. Firstly, communication can be instrumental in providing the accountability that independent central banks should offer (see, for example, Blinder (1998), ch. 3). In addition to this democratic motivation, there is an economic one. As expectations play an important role in economic decision-making, communications by the central bank that influence expectations provide a channel for stabilization of prices and output (see Woodford (2006) or Blinder et al. (2008) for further discussion). Recent empirical work supports this view by showing how central bank communication is an important source of information for economic agents. For instance, Ehrmann and Fratzscher (2007a) find that communications by the Federal Reserve, the Bank of England and the European Central Bank have been an important driver of financial markets. (3)

Still, there is a continuing debate on what would be an optimal communication strategy. There are many open issues: how much information should be shared, how often should it be shared, in which form should it be shared, and how clear should the information be? This article focuses on this last issue. Intuitively, more precise information seems beneficial from a social welfare perspective. However, a number of theoretical papers have argued that restricting transparency could be worth considering. Cukierman and Meltzer (1986) argue that ambiguity enables monetary authorities to generate surprise inflation and stimulate economic activity. The central bank may, therefore, not choose the minimum feasible level of ambiguity, but rather a higher level as this helps in stimulating the economy. Stein (1989) argues that it is not credible for the central bank to announce a target for the exchange rate. One solution may be to make imprecise announcements, for example, by specifying ranges for the exchange rate rather than a single target level. Finally, Morris and Shin (2002, 2005, 2008) have argued that more precise public information may lead to lower social welfare, as it acts as a focal point for economic agents. In their model, economic agents care about the correctness of their actions, but also intend to keep their actions closely in line with the actions of other agents. In this set-up public information serves as a focal point for the beliefs of the public. Morris and Shin show that if the coordination motive is sufficiently strong, and if private information is relatively precise, an increase in precision of public information can actually lead to lower social welfare. The intuition is that agents may place too much weight on the relatively uninformative public signal.

The empirical research on central bank transparency has rapidly grown over the years (see Geraats (2002) or van der Cruijsen and Eijffinger (2007) for an overview). The literature has focused on constructing and using measures of institutional transparency. These indicators focus on various aspects of transparency, such as the publication of a target for inflation, publication of policy deliberations and publications of macroeconomic forecasts. A wel1-known example is the indicator constructed by Eijffinger and Geraats (2006). There has been little attention for the actual quality of the information provided. One exception is the monograph by Fracasso, Genberg, and Wyplosz (2003), which evaluates the quality of inflation reports by 20 central banks. They find that the perceived quality of the writing style negatively correlates with monetary policy surprises--which suggests that clarity indeed reduces uncertainty in financial markets.

Two other related papers focus on Morris and Shin's argument on the precision of public information. Svensson (2006) stresses that the negative welfare effect of more precise public information only occurs if public information has a much lower signal-to-noise ratio than that of private information--which is not very realistic. Ehrmann and Fratzscher (2007b) find some evidence in favor of the Morris and Shin model in their study of communications by the Federal Reserve. Using the degree of dispersion in comments by various central bankers as a measure for precision, they find that the effectiveness of communication depends on the market environment. A statement may enhance predictability when market uncertainty is high, but may reduce it when uncertainty is low. This finding suggests two things: first, the relative precision of communication determines its effect and second, limits to transparency may indeed exist.

Finally, Kohn and Sack (2004) show that various forms of communication by the Federal Reserve influence the volatility of asset prices. The reactions of financial markets to central bank talk indicate that central bank communication conveys important information to market participants. In particular, Kohn and Sack find that testimonies by Alan Greenspan have affected a range of financial prices. In fact, the effects of Greenspan's testimonies are estimated to be larger than those of his speeches. Moreover, the effects of his testimonies extend out beyond the short end of the yield curve, which suggests that the testimonies are perceived to contain more than just information on the short term policy outlook.


The effect of clarity on volatility seems intuitive. Presumably, clearer communication is more easily understood. If so, clarity could effect volatility in financial markets. Blinder (2008), for instance, suggests that clearer communications have higher signal-to-noise ratios. When central bankers communicate clearly, financial markets have less difficulty in interpreting the central bank's message. This leads to less uncertainty, which is reflected in lower volatility. (4)

In this article, it is tested whether clarity and volatility are related and, if so, whether clarity leads to less volatility. To this end, regressions are used in which measures of volatility are the dependent variables, and measures of clarity and a number of control variables are the explanatory variables. To study clarity, this article focuses on readability, which is defined as the ease with which a text can be read. Readability is studied because it is a fundamental precondition for transparency. If it is difficult to read a text, the content is likely to get diffused. There is also an increased likelihood that the reader does not finish reading the text.

There is a long history of research (Flesch 1948; Kibby 1975) that has identified text characteristics, such as lengths of words and sentences, as good predictors of readability. The most important benefit of these readability measures is that they are based on objective elements of the underlying texts. Taking other elements of communication into account through content analysis would introduce a degree of subjectivity into the analysis (for further discussion, see Blinder et al. (2008)). Readability measures have been widely applied in other fields, such as medicine and psychology. An application related to this article is the work of Resche (2004), who uses readability scores in a linguistic analysis of testimonies by Alan Greenspan. Two applications in economics are the works of Laband and Taylor (1992) and Diamond and Levy (1994), who use readability statistics to evaluate the writing style of the economics profession.

Two well-established statistics from the literature on readability are used. The first measure, the Flesch (1948) reading ease score, uses three variables: the number of words, sentences, and syllables used in the text. On the basis of these variables, the Flesch reading ease score is calculated as follows:

(1) 206.835 - 1.015 x (words/sentences) - 84.6 x (syllables/words)

The Flesch score embodies the idea that a higher number of words per sentence, or a higher number of syllables per word, decreases the readability of a text. Texts with Flesch scores above 70 are considered very easy to read, whereas texts which score below 30 are considered very difficult. (5) A drawback of the Flesch statistic is the scaling. For instance, it is not clear how much better a score of 50 is compared to a score of 40. Therefore, the estimated coefficient for the Flesch score is less helpful in assessing the magnitude of the effect of clarity. This problem is remedied by the second statistic: the Flesch-Kincaid grade level (Kincaid et al. 1975), This variable expresses readability as the number of years of education needed to comprehend a text. It is calculated as:

(2) 0.39 x (words/sentences) +11.8 x (syllables/words) - 15.59

We can use the estimated coefficient for the Flesch-Kincaid variable to put the costs or gains of clarity in perspective. This coefficient measures the effect--in terms of volatility--of a decrease in clarity for which an additional year of schooling is needed in order to still comprehend the text.

The readability statistics are applied to the prepared testimony by the Federal Reserve Chairman in the context of the Humphrey-Hawkins (HH) hearings. (6) The basis for the HH hearings is the Full Employment and Balanced Growth Act of 1978, which requires the Federal Reserve to semiannually deliver a report to Congress. This report addresses monetary policy in the context of the current and prospective performance of the U.S. economy. Using the HH testimonies enables me to study developments in clarity of communication by one of the major central banks over a prolonged period of time.

The sample period starts in February 1979 and ends in July 2009. In total, 62 testimonies are analyzed. The testimonies for the period July 1996 to July 2009 were downloaded from the website of the Federal Reserve. (7) For February 1979 to February 1996, the Federal Reserve Archival System for Economic Research (FRASER) at the St. Louis Fed is used. (8) To compute the readability statistics, the software program Flesh 2.0 is used. (9)

Figure 1 shows the Flesch score (top panel) and the Flesch-Kincaid grade level (bottom panel) for each HH testimony between 1979 and 2009. The Flesch score indicates that the HH-testimonies are, in general, difficult to read. Also, over time, the clarity of the testimonies has slowly diminished. Between 1980 and 1990, the Flesch score ranges between 30 and 40. After 1990, it ranges between 26 and 32. The Flesch-Kincaid grade level tells a similar story. In the 1980s, 14-16 years of education were needed to understand the testimonies, while after 1990, at least 15 years were required. (10) To see what drives the decline in readability, Figure 2 plots the development of the two components of the readability statistics: the number of words per sentence (dashed line, right axis) and the number of syllables per word (solid line, left axis). The decline in readability turns out to be driven by an increase of the number of syllables per word. In contrast, the number of words per sentence has remained stable over time.



The level of readability is not the main focus of this article. However, it is interesting to note that the decline of clarity is in stark contrast to the conventional wisdom that Federal Reserve transparency has increased over the years (Blinder et al., 2008). This divergence shows that transparency does not automatically translate into clarity. It will be interesting to further explore the relationship between clarity and transparency in future work.

Using the readability statistics, effects of clarity on volatility for various financial variables are studied. Table 1 lists these variables. Data on interest rates, exchange rates, and the stock market are used. Volatility is measured as the standard deviation of daily changes in yields (for the bond data) or returns (for the exchange rate and stock indices). The event window starts on the day of the testimony and includes ten business days. Figure 3 displays developments in volatility for a number of financial instruments. In all cases--also those not displayed in the figure--volatility is high in the early 1980s, and shows a decline over the period 1990-2007, before often rising again in the years 2008 and 2009. (11)
Overview of Dependent Variables

Data Series Source

Federal funds rate Federal Reserve Statistical Release H. 15
Treasury yield 3 months Federal Reserve Statistical Release H. 15
Treasury yield 1 year Federal Reserve Statistical Release H. 15
Treasury yield 2 year Federal Reserve Statistical Release H. 15
Treasury yield 3 year Federal Reserve Statistical Release H. 15
Treasury yield 5 year Federal Reserve Statistical Release H. 15
Treasury yield 10 year Federal Reserve Statistical Release H. 15
S/DM Federal Reserve Statistical Release H. 10
$/Euro Federal Reserve Statistical Release H. 10
Dow Jones Ind. Avg. Datastream (close price)

Notes: All Treasury yields are based on actively traded
issues adjusted to constant maturities. Data accessed through: URL last accessed on
October 20, 2009.

For three reasons, the baseline analysis used ten-day event windows. A first consideration is that one needs to define sufficiently long event windows in order to have a sensible estimate of volatility. Given that daily data are used, at least a few days are needed to compute the standard deviation of yield changes and returns.

Second, it is not clear, a priori, how long the information in the testimonies stays relevant for market participants. In general, financial markets quickly incorporate new information. However, if there is uncertainty about the nature of the information, it may take longer for the markets to decide on the message of the central bank. On the day of the testimony, there may be live coverage (for example over C-SPAN), and reports will be made through news-wires such as Reuters or Bloomberg. By now--but not during the 1980s and 1990s--the text of the testimony will also be published on the internet. Later during the day, there will be coverage of the testimony through television and radio. On the day following the testimony, newspapers will publish stories relating and interpreting the testimony. Finally, there will be discussion of the testimony in various fora, such as magazines, web-logs, and analyst meetings, during a number of days following the testimony. (12)


Thirdly, the policy perspective is important. From an econometric point of view, identifying the effects of clarity would most readily be performed using intraday data. (13) However, if the effects of clarity are beneficial, one would like them to persist for as long as possible. The literature shows different findings on the persistence of the effects of communication. Regarding exchange rates, for instance, Jansen and De Haan (2007) find that verbal interventions had only very short-run effects, whereas Fratzscher (2009) finds that communication by the G7 was able to move exchange rates in the intended direction for horizons up to three months after meetings.

The bottom line is that the ideal length of the event window is not a priori clear, but could extend over a number of trading days. The appendix shows whether the effects of clarity are robust to varying the length of the event window. One remaining issue is whether to include the day of the testimony itself in the event window. In this article, event windows that include the day of the testimony are used. The results are, however, robust to excluding these days.

To identify the effects of clarity, the natural logarithm of the volatility measures are regressed on the measures of readability and a number of control variables. If clarity has the expected effect on volatility, the estimated coefficient is negative when using the Flesch score and positive for the Flesch-Kincaid grade level. Four categories of control variables are used. First, it is common in event studies to control for pre-event volatility. Often, researchers will do so by using the ratio of post-event and pre-event volatility as the dependent variable (see Clayton, Hartzell, and Rosenberg (2005) or Dubofsky (1991)). However, by using this ratio, one implicitly assumes the coefficient for pre-event volatility equals one. Here, a more flexible approach by including pre-event volatility as a separate control variable is used. The length of the pre-event window is always equal to the length of the post-event window. Second, volatility and clarity may well depend on business cycle developments. Three macroeconomic variables are included: growth in real gross domestic product (GDP), inflation, and changes in the money supply. (14) The inclusion of GDP and inflation is motivated by the Federal Reserve's mandate, while the money supply is included because it figured prominently in the Federal Reserve's monetary policy strategy for a large part of the sample period. For GDP, the reported quarterly growth of the quarter in which the testimony took place is used. For inflation, the year-on-year change of the consumer price index (CPI) of the month of the testimony is used, while for the money supply, monthly growth in M2 is used. (15) Revised macroeconomic data are used, but the appendix shows that the results are robust to using real-time data.

Third, the content of the Humphrey-Hawkins hearings is taken into account. Kohn and Sack (2004) have shown how new information presented at Humphrey-Hawkins testimonies influences volatility in financial markets. Apart from the news itself, the type of news may be important. For instance, bad news about the economy might lead policy makers to make more effort to explain their efforts. As objective measures for communication are preferred, this article uses the projections for economic growth and inflation which are given in each testimony. In the end, the projections for the current year are used, as these are available for the whole sample, while projections for the next year are not. The projections are communicated as a range, rather than a point estimate. (16) To gauge the economic outlook of the FOMC, the average of the top and bottom end of the range is used. In addition, the difference between the top and bottom of the range is used as a measure of the uncertainty surrounding the economic outlook. As uncertainty may translate in additional volatility, a positive coefficient for these uncertainty measures is expected. As a robustness check, the news contained in the testimonies was also measured with a subjective measure by constructing an indicator of the "monetary policy inclination" and the "economic outlook" as in Ehrmann and Fratzscher (2007a). However, these variables were never significant in the regressions.

Fourthly, an important element is the role of the Chairman. The Humphrey-Hawkins testimonies have been given by four different individuals: William Miller (1979), Paul Volcker (1980-1987), Alan Greenspan (1988-2005), and Ben Bernanke (since 2006). There may well be important differences in communication styles which should be controlled for. Therefore, binary dummies are included for the various Chairmen. Finally, in order to capture potential differences in volatility between February and July, a binary dummy is included that equals one for the first testimony of each year.


Table 2 shows the estimation results when the Flesch score is used, while Table 3 has results for the Flesch-Kincaid grade level. The explanatory variables are listed in the first column. The remaining columns have results for various financial instruments--ranging from the Federal funds rate to the Dow Jones Industrial Average. There are two important results. First, clarity has mattered, but not for each financial instrument. There are four cases in which the full sample results show significant results for clarity: the 2-year bond, the 3-year bond and the Dow Jones index (using both readability measures) and the euro-dollar exchange rate (for the Flesch-Kincaid measure). The fact that significant results are mainly found for the short to medium end of the yield curve is not surprising. In general, near-term rates will be more dependent on the expectations for future monetary policy actions than longer rates. For instance, Ehrmann and Fratzscher (2007a) present empirical evidence that communication on the monetary policy inclination and the economic outlook mainly affects near-term interest rates--although the results vary somewhat between the various central banks which they study. What is perhaps surprising is that there is no role for clarity with respect to volatility of very-short run instruments, such as the federal funds rate or the three-month bond.
Effects of Readability: Results Using Flesch Reading Ease Score

 Fed Trs. Trs. Trs. Trs. Trs. Trs.
 funds 3m. ly. 2y. 3y. 5y. l0y.

Flesch -0.01 -0.02 -0.01 -0.04 -0.03 -0.02 -0.02
 *** **

[[sigma].sup.pre] 0.72 0.39 0.34 0.40 0.56 0.43 0.55
 *** ** ** *** *** *** ***

GDP -0.19 -0.11 -0.11 -0.09 -0.03 -0.02 0.00
 ** *

CPI -0.07 0.04 -0.03 -0.02 -0.03 -0.00 -0.01

M2 0.06 0.29 0.23 0.10 0.18 0.04 0.05

Ec. Growth (f) 0.08 -0.02 0.04 0.04 -0.00 -0.00 -0.01

Inflation (f) 0.33 0.20 0.24 0.16 0.13 0.04 0.04
 ** * *

Ec. Growth (u) -0.05 -0.03 -0.03 0.01 -0.03 0.13 -0.02

Inflation (u) -0.37 -0.29 -0.16 -0.05 0.07 0.03 0.14

Miller -2.31 -2.86 -0.81 -0.83 -1.46 -0.98
 ** *** **

Volcker -1.46 -1.81 -2.28 -1.11 -0.75 -1.10 -0.71
 * ** *** ** **

Greenspan -1.38 -2.22 -2.37 -1.12 -0.63 -1.13 -0.74
 ** *** *** ** ***

Bernanke -1.44 -2.15 -2.22 -1.02 -0.62 -1.13 -0.82
 * ** *** ** *** *

February 0.20 0.16 0.21 0.16 0.05 0.03 0.02
 ** *

Adj. [R.sub.2] 0.73 0.65 0.68 0.60 0.54 0.50 0.53

JB 3.60 0.20 0.84 2.65 0.17 0.89 2.07

LM (2) 0.30 0.26 0.47 0.46 2.26 0.04 1.08

ADF -8.24 -6.44 -8.40 -8.70 -9.23 -7.81 -8.54
 *** *** *** *** *** ** ***

 $-DM $-Euro DJIA

Flesch 0.01 -0.04 -0.02

[[sigma].sup.pre] 0.06 0.40 * 0.52

GDP -0.01 -0.02 -0.06

CPI 0.02 -0.04 -0.01

M2 0.08 -0.04 0.25

Ec. Growth (f) 0.06 0.11 0.10

Inflation (f) 0.03 -0.03 0.02

Ec. Growth (u) -0.12 0.47 -0.04

Inflation (u) 0.21 0.09 0.32

Miller -6.77 -2.03
 *** **

Volcker -5.60 -2.29
 *** **

Greenspan -5.57 -2.33 -2.14
 *** * **

Bernanke -2.52 -1.85
 * **

February -0.17 -0.10 0.02

Adj. [R.sub.2] -0.07 0.35 0.45

JB 1.17 0.57 1.08

LM (2) 0.44 0.91 0.49

ADF -5.38 -3.41 -7.49
 *** ** ***

Notes: Dependent variables are measured as natural logarithms
Of volatility measured over ten-day window.
[[sigma].sup.pre] denotes pre-event volatility. GDP, CPI, and
M2 are macroeconomic controls. Ec. growth and inflation are
based on projections detailed in the Humphrey-Hawkins testimonies,
where (f) denotes forecast, and (u) denotes uncertainty. February
is a binary dummy for the first testimony of each year. Trs. denotes
constant-maturity Treasury yields. JB denotes Jarque-Bera statistic.
LM(2) denotes F-statistic for Breusch-Godfrey test. ADF
denotes r-statistic for augmented Dickey-Fuller test on the residuals.
All estimations based on least-squares regression with Newey-West
corrected standard errors.
*/**/*** denotes significance at the 10/5/1% level.
Effects of Readability: Results using Flesch-Kincaid Grade Level

 Fed Trs. Trs. Trs. Trs. Trs. Trs.
 funds 3m. ly. 2y. 3y. 5y. l0y.

Flesch-Kincaid 0.02 0.10 0.08 0.14 0.12 0.05 0.07
 ** *

[[sigma].sup.pre] 0.72 0.38 0.35 0.42 0.59 0.44 0.56
 *** ** ** *** *** *** ***

GDP -0.18 -0.09 -0.10* -0.06 -0.01 -0.01 0.02

CPI -0.07 0.04 -0.03 -0.01 -0.02 0.00 -0.00

M2 0.05 0.29 0.22 0.08 0.16 0.03 0.04

Ec. Growth (f) 0.08 -0.02 0.04 0.03 -0.01 -0.01 -0.02

Inflation (u) 0.32 0.21 0.24 * 0.14 0.11 0.02 0.02

Ec. Growth (u) -0.04 -0.04 -0.04 0.03 -0.03 0.15 -0.01

Inflation (u) -0.37 -0.30 -0.17 -0.07 0.04 0.02 0.12

Miller -2.87 -4.51 -4.24 -3.62 -2.93 -2.77
 * *** *** ** ** **

Volcker -1.98 -4.03 -3.90 -4.35 -3.40 -2.45 -2.40
 *** *** *** ** * **

Greenspan -1.89 _4.42 -3.96 -4.31 -3.23 -2.45 -2.40
 *** *** *** ** * **

Bernanke -1.95 -4.35 -3.82 -4.19 -3.21 -2.44 -2.48
 *** *** *** ** * **

February 0.21 0.19 0.23 0.19 0.08 0.04 0.03
 ** **

Adj. [R.sup.2] 0.73 0.65 0.69 0.58 0.53 0.49 0.52

JB 3.51 0.10 0.68 2.85 0.37 0.86 2.57

LM (2) 0.31 0.39 0.54 0.47 2.49 0.05 1.02

ADF -8.27 -6.37 -8.37 -8.72 -9.16 -7.80 -8.42
 *** *** *** *** *** *** ***

 $-DM $-Euro DJIA

Flesch-Kincaid 0.00 0.22 * 0.10

[[sigma].sup.pre] 0.08 0.41 * 0.54

GDP -0.02 0.02 -0.04

CPI 0.01 -0.08 -0.01

M2 0.10 0.05 0.23

Ec. Growth (f) 0.06 0.13 0.10

Inflation (u) 0.06 0.0) 0.03

Ec. Growth (u) -0.16 0.38 -0.05

Inflation (u) 0.20 0.13 0.30

Miller -6.12 -4.26
 *** ***

Volcker -5.11 -4.44
 *** ***

Greenspan -5.10 -6.93 -4.27
 *** ** ***

Bernanke -7.07 -3.98
 ** ***

February -0.16 -0.04 0.05

Adj. [R.sup.2] -0.08 0.45 0.46

JB 1.11 0.22 2.70

LM (2) 0.39 1.48 0.45

ADF -5.28 -4.39 -7.47
 *** *** ***

Notes: Dependent variables are measured as natural logarithms of
volatility measured over ten-day window. [[sigma].sup.pre] denotes
pre-event volatility. GDP, CPI, and M2 are macroeeonomic controls.
Ec. growth and inflation are based on projections detailed in the
Humphrey-Hawkins testimonies, where (f) denotes forecast, and (u)
denotes uncertainty. February is a binary dummy for the first
testimony of each year Trs. denotes constant-maturity Treasury yields.
JB denotes Jarque-Bera statistic. LM (2) denotes F-statistic for
Breusch-Godfrey test. ADF denotes t-statistic for augmented
Dickey-Fuller test on the residuals. All estimations based on
least-squares regression with Newey-West corrected standard errors.
*/**/*** denotes significance at the 10/5/1% level.

The signs of the significant parameters in Tables 2 and 3 introduce the key result of this article. When clarity matters, the results are unequivocal: clarity diminishes volatility. For the Flesch score, significant parameters are negative, meaning that a higher degree of readability means lower volatility. For the Flesch-Kincaid grade level, the significant parameter is positive, indicating that a higher number of years of education needed to comprehend a text coincides with a higher degree of volatility. Perhaps one may worry that the results for readability are driven by the trends in readability and volatility. However, the macroeconomic control variables also display trend behavior, and this seems sufficient to account for the trend in volatility. A separate trend term in the regressions was included as a robustness check, but the trend was not significant. More importantly, if trend behavior drives the results, this should affect the coefficient for readability differently. As both volatility and the Flesch statistic show a downward trend, the coefficient would have been positive rather than negative. In case of the Flesch-Kincaid statistic, the coefficient would have been negative instead of positive. (17)

As expected, the coefficient for the pre-event volatility variables is often highly significant and positive. Using Wald tests, one can indeed always reject the hypothesis that this coefficient equals one which justifies using pre-event volatility as a separate control variable. Concerning the macroeconomic variables, the significant cases are concentrated at the short end of the yield curve. For the four variables measuring the content of the Humphrey-Hawkins testimonies, especially the coefficients for expected inflation are significant. Again, they are concentrated at the short end of the yield curve. Finally, the Chairman dummies are highly significant, while the dummy for the February testimony is in some cases significantly larger than zero. (18)


The full sample analysis assumes a constant effect of clarity. This assumption may be too rigid. Firstly, Federal Reserve policy itself has been subject to change. For example, over time, monetary aggregates have become less important in the Federal Reserve's monetary policy strategy. Also, there have been important changes in Federal Reserve transparency. For instance, changes in the federal funds rate target are only announced since 1994. Also, only since 1999, there has been a statement after each meeting of the FOMC. These changes in transparency may have changed the relative impact of the Humphrey-Hawkins testimonies. Third, the use of technology in financial markets is much more widespread. As a result, the speed of dissemination of information has greatly increased. By now, electronic trading platforms have largely replaced brokers. Also, information systems provided by companies such as Reuters and Bloomberg have changed the way in which information reaches financial market participants. Together these two transformations may have changed the manner in which central bank communication is processed. Finally, as noted, the testimonies have been given by four different Chairmen, who perhaps have had very different rhetorical styles.

Overall, there is sufficient reason to flesh out the effect of clarity in different periods. To this end, rolling-window regressions are estimated. The first subsample uses the first 15 years of the full sample: 1979-1993. For the subsequent windows, the start and end-points are moved forward by 1 year. The results are shown in Figure 4. (19) The top panel has results for the Flesch score, the bottom panel has results for the Flesch-Kincaid grade level. The graphs show the upper bound (in case of the Flesch score) and the lower bound (in case of the Flesch-Kincaid grade level) of the 95% confidence interval for the coefficient for readability. There are two striking features. Most importantly, the message that readability diminishes volatility is borne out in both panels. When the readability measures are significant, the signs of the coefficients point out this conclusion. However, it is also apparent that the effects of clarity have greatly varied over time. At first, the coefficient for readability--for the Flesch as well as the Flesch-Kincaid measure--is significant for the Dow Jones index, but not for interest rates. For subsamples starting after the mid-1980s, this pattern is reversed. In particular, significant estimates are found in case of the 2-year and 3-year maturity. For the last few subsamples which are analyzed, however, the significant effect of clarity has largely disappeared.


The patterns in Figure 4 could be related to the role of the Federal Reserve Chairman. There may well be important differences in communication styles between Chairman. For instance, the conventional wisdom is that communications by Alan Greenspan were relatively inaccessible. Greenspan himself suggested as much by stating that as a central banker he had "learned to mumble with great incoherence." (20) To analyze a possible Chairman-effect, regressions for sub-samples defined by Chairmanship are estimated. Table 4 shows the results. Rows 1 and 2 show estimates for the effect of readability when using only testimonies by Greenspan. Rows 3 and 4 present results for the Volcker period, while rows 5 and 6 combine testimonies by Miller and Volcker. Rows 7 and 8 present results for all non-Greenspan observations. (21)
Was the Greenspan Term Different?

 Fed Trs. Trs. Trs. Trs. Trs. Trs.
 funds 3m. ly. 2y. 3y. 5y. l0y.


(1) Flesch -0.05 -0.00 -0.04 -0.06 -0.04 -0.03 -0.03
 * * *** **

(2) Flesch-Kincaid 0.23 0.09 0.23 0.29 0.19 0.09 0.12
 * ** ** *


(3) Flesch 0.08 0.10 0.03 -0.04 -0.06 -0.03 -0.04

(4) Flesch-Kincaid -0.37 -0.52 -0.03 0.24 0.29 0.17 0.17
 * *


(5) Flesch 0.08 -- 0.02 -0.04 -0.06 -0.03 -0.03

(6) Flesch-Kincaid -0.33 -- -0.03 0.22 0.29 0.17 0.17


(7) Flesch 0.04 -0.03 -0.01 -0.02 -0.02 -0.02 -0.02

(8) Flesch-Kincaid -0.15 0.15 0.05 0.08 0.08 0.06 0.06



(1) Flesch 0.01 -0.03

(2) Flesch-Kincaid 0.01 0.20


(3) Flesch 0.06 -0.03

(4) Flesch-Kincaid -0.21 0.12


(5) Flesch 0.05 -0.02

(6) Flesch-Kincaid -0.12 0.04


(7) Flesch -- -0.02

(8) Flesch-Kincaid -- 0.10

Notes: Estimates of the coefficient for the readability measures. Trs.
denotes constant-maturity Treasury yields. All estimations based on
least-squares regression with Newey-West corrected standard errors.
*/**/*** denotes significance at the 10/5/1% level.


Firstly, there are a number of cases in which the coefficient for readability is significant. Again, the signs of the significant coefficients indicate that clarity diminishes volatility. (22) Using the coefficients for the Flesch-Kincaid grade level, we can now also put the benefits of clarity into perspective. The significant coefficients in Table 4 range up to 0.29. Suppose that a testimony would change in such a way, that an additional year of schooling is necessary to still sufficiently comprehend its content. In that case, the conditional effect on volatility could be an increase by as much as 29%.

Secondly, there are indeed indications of Chairman effects. For the Greenspan sample, more significant effects are found than for the other two samples, which suggests that clarity mattered especially during his period as Chairman. In contrast to the full sample results, for Greenspan testimonies, significant effects for the shorter end of the yield curve are also found. For instance, volatility of both the federal funds rate and the 1-year bond are affected by clarity. Still, we have to take into account that a larger number of observations for Greenspan are used than for the other three Chairmen, which helps in identifying possible effects of clarity. Also, significant effects are not exclusively because of Greenspan. During the Volcker Chairmanship, there are effects on 3-year interest rates and the Dow Jones index. So, although clarity mattered in particular during Alan Greenspan's term, it has also been important during earlier periods.


Central bank transparency has been debated for many years now. This article presents an empirical analysis of an important precondition for transparency: clarity of communication. Clarity does not turn out to be a panacea, as there is no constant and widespread effect on volatility. Still, in a number of cases, there is a significant effect of clarity. In those cases, the direction of the effect is intuitive: greater clarity coincides with less volatility in financial markets.

This article stresses the importance of transparent communication by monetary authorities. A negative relationship between clarity and volatility has clear policy implications. Over the last two decades, many central banks have undertaken important steps toward greater transparency. As a result, the predictability of monetary policy has greatly increased (Blinder et al., 2008). This article suggests that central banks could reap further benefits--in terms of reduced volatility--by paying close attention to the clarity of their communication. Given that the marginal cost of clarity is not prohibitively large--in fact, it is mainly the opportunity cost of staff doing additional editing of central bank communications--the gains of clarity seem well worth pursuing.

Future research could extend the analysis in various ways. Of course, the readability measures could be applied to communications by other central banks. Also, other channels of communication by the Federal Reserve could be taken into account. Thirdly, it would be interesting to flesh out the short-run effects of clarity using intra-day data. Perhaps the most interesting route will be to explore the effects of clarity on the predictability of monetary policy decisions. These issues are left for future work.


First, the length of the event window over which volatility is computed is varied. Table A1 shows estimated coefficients for readability for various window lengths between 3 and 15 days. Column 1 shows results for the Fiesch score, while column 2 has results for the Flesch-Kincaid grade level. On the basis of the signs of significant parameters, the conclusion remains that clarity diminishes volatility. In terms of significance, there is quite some variation. In general, the results are robust to shorter and longer event windows. Finally, it must be noted that outliers sometimes influence the results when shortening the window length. Overall, the most robust case is the 2-year interest rate.
Robustness: Length of Event Window

 (1) Flesch-
Instrument Window Flesch Kincaid

Trs. 2y. 15 clays -0.02 0.08
 13 days -0.03 ** 0.11 *
 9 days -0.04 *** 0.14 **
 8 days -0.04 *** 0.14 ***
 7 days -0.01 0.01
 5 days -0.01 0.03
 3 days -0.04 ** 0.12
Trs. 3y. 15 days -0.02 0.08
 13 days -0.02 * 0.09
 9 days [dagger] -0.03 * 0.10
 8 days [dagger] -0.03 ** 0.09
 7 days 0.01 -0.03
 5 days 0.01 -0.10
 3 days -0.02 -0.02
$-Euro 15 days -0.05 ** 0.18 **
 13 days -0.02 0.15 *
 9 days -0.04 0.24 *
 8 days -0.02 0.26 *
 7 days -0.02 0.25
 5 days -0.06 0.42 **
 3 days -0.03 0.27 *
DJIA 15 days -0.02 * 0.12 *
 13 days -0.02 * 0.12 **
 9 days -0.02 0.08
 8 days -0.02 ** 0.13 **
 7 days -0.02 0.09
 5 days -0.01 0.07
 3 days -0.01 0.06

Notes: Estimates of coefficient for readability when varying the event
window length. The significant results in the baseline specification
are focused (Tables 2 and 3). For other variables, the parameter
estimates for readability remain insignificant for various event
windows. Trs. denotes constant-maturity Treasury yields. The [dagger]
denotes that a dummy is used to correct for one outlier. All
estimations based on least-squares regression with Newey-West
standard errors.
*/**/*** denotes significance at the 10/5/1% level.

Second, the baseline model conditions on pre-event volatility. This variable can be interpreted as a measure of the level of uncertainty in financial markets before Humphrey-Hawkins testimonies. Other measures could also proxy for macroeconomic uncertainty. Two alternative measures derived from the Livingston survey and Consensus Forecasts are used. As a proxy for uncertainty, the standard deviation of the responses to the last survey before a testimony is used. For the Livingston survey, inflation expectations are used, and for Consensus expectations on three-months interest rates are used. The Livingston data are available for the full sample period, the Consensus data for 1990-2009. (23) The results are in Table A2. Columns 1 and 2 show the coefficients for the readability statistics, column 3 shows the coefficient for the Livingston proxy. and column 4 shows the coefficient when using the Consensus proxy. Most important, based on the sign of significant parameters, the conclusion regarding the role of clarity remains unchanged. The results for the Consensus variable are more according to intuition than those for the Livingston survey variable. Estimations with the Livingston and Consensus proxies included as additional variables in the baseline model are also run. This left the conclusions regarding the effects of clarity unchanged.
Robustness: Other Proxies for Uncertainty

 (1) (2) (3) (4)
Instrument Flesch Flesch-Kincaid Livingston Consensus

Fed funds (1) -0.01 -0.74 *

 (2) -0.03 -0.74 *

 (3) -0.05 -0.10

 (4) 0.27 -0.04

Trs. 3m. (1) -0.02 -0.15

 (2) 0.11 -0.12

 (3) -0.02 0.21

 (4) 0.25 * 0.27

Trs. ly. (1) -0.02 -0.25

 (2) 0.05 -0.23

 (3) -0.03 0.70 ***

 (4) 0.19 ** 0.74 ***

Trs. 2 y. (1) -0.03 -0.10

 (2) 0.10 -0.05

 (3) -0.05 0.61 ***

 (4) 0.19 ** 0.74 ***

Trs. 3y. (1) -0.03 -0.03

 (2) 0.08 0.01

 (3) -0.07 0.57 ***
 [dagger] ***

 (4) 0.27 *** 0.54 ***

Trs. 5y. (1) -0.02 -0.01

 (2) 0.05 0.02

 (3) -0.05 0.36 ***
 [dagger] ***

 (4) 0.18 * 0.50 **

Trs. l0y. (1) -0.03 -0.05

 (2) 0.10 -0.01

 (3) -0.03 0.33

 (4) 0.18 ** 0.36 *

$-Euro (1) -0.04 -0.54 **

 (2) 0.17 -0.41 **

 (3) -0.03 -0.22

 (4) 0.22 * -0.23

DJIA (1) -0.03 -0.12
 [dagger] **

 (2) 0.13 * -0.09

 (3) -0.02 0.24

 (4) 0.20 ** 0.29

Notes: Estimates when pre-event volatility is replaced by other
proxies for uncertainty. The Consensus data is available for the period
1990-2009. Trs. denotes constant-maturity Treasury yields. The [dagger]
denotes that dummies are used to correct for outliers. All estimations
based on least-squares regression with Newey-West standard errors.
*/**/*** denotes significance at the 10/5/1% level.

Thirdly, all regressions are re-estimated using detrended data. The data are detrended using a Hodrick-Prescott filter which is more flexible than a linear trend. As semi-annual data are used, the smoothing parameter is set at 400. Table A3 shows that the conclusions regarding clarity are not affected. Again, the coefficient for clarity is significant for the 2-year rate, the 3-year rate, the euro-dollar rate and the Dow Jones index. The signs of the coefficients again show how clarity diminishes volatility.
Robustness; De-trended Data

 Flesch Flesch-Kincaid

Fed funds 0.00 -0.06
Trs. 3m. -0.02 0.10
Trs. ly. -0.01 0.06
Trs. 2y. -0.03 ** 0.12 *
Trs. 3y. -0.03 * 0.11
Trs. 5y. -0.02 0.04
Trs. l0y. -0.02 0.06
$-DM 0,01 -0.01
$-Euro -0.03 * 0.15
DJIA -0.02 ** 0.10 *

Notes: Estimates for readability statistics when data is de-trended.
Trs. denotes constant-maturity Treasury yields. All estimations based
on least-squares regression with Newey-West standard errors.
*/**/*** denotes significance at the 10/5/1% level.

Finally, one may question the use of revised macroeconomic data. Volatility is determined by the information available to economic agents at a given point in time. Therefore, it could be advisable to use real-time macroeconomic data. At the same time, the revisions in GDP, CPI, and M2 are perhaps not so large as to overturn the estimation results. Indeed, this turns out to be the case. Realtime data on GDP, CPI, and M2 are downloaded from the Philadelphia Federal Reserve website. (24) First, a visual inspection showed that revised and real-time data show similar developments over time. Second, all regressions are re-estimated using real-time data. The coefficients for readability are displayed in Table A4. The coefficients are similar to the results with revised macro data. Most importantly, the conclusion remains that clarity diminishes volatility.
Robustness: Real-time Macroeconomic Data

 Flesch Flesch-Kincaid

Fed funds 0.00 0.07
Trs. 3m. -0.02 0.06
Trs. ly. -0.01 0.07
Trs. 2y. -0.04 *** 0.16 **
Trs. 3y. -0.03 ** 0.12 *
Trs. 5y. -0.02 0.07
Trs. l0y. -0.02 * 0.09
S-DM 0.02 -0.01
S-Euro -0.05 *** 0.23 **
DJIA -0.02 * 0.09

Notes: Estimates for readability statistics when real-time GDP,
CPI, and M2 data is used. Trs. denotes constant-maturity Treasury
yields. All estimations based on least-squares regression with
Newey-West standard errors.
*/**/*** denotes significance at the 10/5/1% level.

(1.) One notable exception is the monograph by Fracasso, Genberg, and Wyplosz (2003) which evaluates inflation reports of 20 central banks.

(2.) Examples include the analysis of readability of informed-consent forms in medicine (Paasche-Orlow, Taylor, and Brancati 2003), the clarity of psychological reports (Smith Harvey 1997), the quality of annual reports in accounting (Clatworthy and Jones 2001) and the readability of abstracts in reading research (Hartley and Sydes 1997).

(3.) The literature has so far focused on communication with financial markets. Communication with the general public is still highly under-researched. It would be interesting to study the role of clarity also in that context.

(4.) Of course, clarity can apply to many elements of communication. For example, we could question whether the objective of monetary policy is "clearly" defined, or whether the monetary policy strategy is properly explained. For a further discussion on clarity, see Winkler (2002).

(5.) The Fiesch formula is the result of a regression in which the dependent variable is the average grade of children who could answer 75% of multiple choice questions based on short texts correctly. Technically, a Fiesch score of 100 indicates reading matter that is understandable for persons who have completed fourth grade. See Fiesch (1948, pp. 224-225) for more details.

(6.) The testimony is followed by a question-and-answer session. A careful reading of newspaper reports on HH testimonies suggests that the prepared testimony often received more attention than the Q&A session. As the Q&A also proceeds in a nonstandardized fashion, it is excluded from analysis. Between 1975 and 1978, testimonies on monetary policy were given under the House Concurrent Resolution 133 and the Federal Reserve Reform Act. These testimonies are not analyzed in this article.

(7.) Source:, last accessed in October 2009.

(8.) Source:, last accessed in February 2008. The testimony is given to the House and the Senate. Often, the texts are similar. As in Kohn and Sack (2004), only the first testimony is used.

(9.) The program is available at The scores from Flesh were cross-checked with scores computed using Word 2003. There were no statistically significant differences.

(10.) Given this trend, augmented Dickey-Fuller tests are used to investigate whether readability can be treated as I(0). The test equation includes a constant and lag lengths were chosen on the basis of the Schwartz criterion. The t-statistic equals -3.09 (p = 0.03) for the Flesch variable, for the Flesch-Kincaid, it equals -2.73 (p = 0.07). To check robustness of the baseline results, equations with readability statistics in first differences, with qualitatively similar results are also estimated.

(11.) ADF tests indicate that the volatility measures are stationary. The case of the euro/dollar rate is questionable, but here the short sample is problematic in using the ADF-test.

(12.) By now, through advances in information technology, it is easier for market participants to watch live transmission of Humphrey-Hawkins testimonies. This raises the issue of differences between clarity of spoken and written communications. However, it seems not unreasonable to assume that, if a text is difficult to read, it will also be difficult to listen to the text being read.

(13.) Data availability concerns would greatly reduce the number of testimonies that could be analyzed. High-frequency data are especially difficult to obtain for the earlier period of the sample.

(14.) Data were downloaded from the St. Louis Federal Reserve website. Source: last accessed in October 2009.

(15.) Using lags of these macroeconomic variables did not affect the estimation results.

(16.) Usually, both the full range and what is called the "central tendency" of the projections by individual Federal Open Market Committee (FOMC) members are given. As the latter receives most emphasis in the testimonies, the central tendency is used.

(17.) As a robustness check, all variables were detrended, and the regressions were re-estimated. The conclusions regarding clarity are unchanged. The Appendix reports these results.

(18.) The Appendix reports various sensitivity results.

(19.) The figure only displays selected results to facilitate readability.

(20.) The statement was made in testimony to a Senate Sub-Committee in 1987. See Geraats (2007) for the full quote.

(21.) The Bernanke sample has so far too little observations to be analyzed separately.

(22.) There is one exception for Volcker and the federal funds rate. Overall, this is the only counterintuitive result for the effect of clarity in this article.

(23.) See also Ehrmann and Fratzscher (2007b), who use similar variables as a proxy for private information in their empirical analysis of Morris and Shin (2002).

(24.) Source:, URL last accessed in October 2009.


ADF: Augmented Dickey-Fuller Test

CPI: Consumer Price Index

FOMC: Federal Open Market Committee

FRASER: Federal Reserve Archival System for Economic Research

GDP: Gross Domestic Product

HH: Humphrey-Hawkins

JB: Jarque-Bera Statistic



Blinder, A. S. Central Banking in Theory and Practice. Cambridge, MA, and London: The MIT Press, 1998.

--. "Central Bank Communication and the Financial Markets." Paper presented at Riksbank conference on 'Refining Monetary Policy: Transparency and Real Stability. 56 September 2008. Available at:

Blinder, A. S., M. Ehrmann, M. Fratzscher, J. De Haan, and D. Jansen. "Central Bank Communication and Monetary Policy: A Survey of Theory and Evidence." Journal of Economic Literature, 46(4), 2008. 908-943.

Clatworthy. M., and M. J. Jones. "The Effect of Thematic Structure on the Variability of Annual Report Readability." Accounting, Auditing and Accountability Journal, 14(3), 2001, 311-326.

Clayton, M. J., J. C. Hartzell, and J. V. Rosenberg. "The Impact of CEO Turnover on Equity Volatility." Journal of Business, 78(5). 2005. 1779-1808.

van der Cruijsen, C. A. B., and S. C. W. Eijffinger. "The Economic Impact of Central Bank Transparency: A Survey." Center for Economic Policy Research Discussion Paper 6070, 2007.

Cukierman, A., and A. H. Meltzer. "A Theory of Ambiguity, Credibility and Inflation under Discretion and Asymmetric Information." Econometrica, 54(5), 1986, 1099-1128.

Diamond, A. M., and D. M. Levy. "The Metrics of Style: Adam Smith Teaches Efficient Rhetoric." Economic Inquiry, 32(1), 1994, 138-145.

Dubofsky, D. A. "Volatility Increases Subsequent to NYSE and AMEX Stock Splits." Journal of Finance, 46(1), 1991, 421-431.

Ehrmann, M., and M. Fratzscher. "Communication by Central Bank Committee Members: Different Strategies, Same Effectiveness?" Journal of Money, Credit and Banking, 39(23), 2007a. 509-541.

--. "Social Value of Public Information: Testing the Limits to Transparency." European Central Bank Working Paper 821, 2007b.

Eijffinger, S. C. W., and P. M. Geraats. "How Transparent Are Central Banks?" European Journal of Political Economy, 22(1), 2006, 1-21.

Flesch, R. "A New Readability Yardstick." Journal of Applied Psychology, 32(3), 1948, 221-233.

Fracasso, A., H. Genberg, and C. Wyplosz. "How do Central Banks Write? An Evaluation of Inflation Reports by Inflation Targeting Central Banks." Geneva Reports on the World Economy Special Report 2. Geneva/London: International Center for Monetary and Banking Studies/Centre for Economic Policy Research, 2003.

Fratzscher, M. "How Successful is the G7 in Managing Exchange Rates?" Journal of International Economics, 79(l), 2009, 78-88.

Geraats, P. M. "The Mystique of Central Bank Speak." International Journal of Central Banking, 3(1), 2007, 37-80.

-- "Central Bank Transparency." Economic Journal, 112(483), 2002, F532-F565.

Hartley, J., and M. Sydes. "Are Structured Abstracts Easier to Read than Traditional Ones?" Journal of Research in Reading, 20(2), 1997, 122-136.

Jansen, D., and J. De Haan. "Were verbal efforts to support the euro effective? A high-frequency analysis of ECB statements." European Journal of Political Economy, 23(1), 2007, 245-259.

Kibby, M. W. "The Proper Study of Readability: A Reaction to Carver's 'Measuring Prose Difficulty Using the Rauding Scale."' Reading Research Quarterly, 11(4), 1975, 686-705.

Kincaid. J. P., R. P. Fishburne, R. L. Rogers, and B. S. Chissom. "Derivation of New Readability Formulas (Automated Readability Index, Fog Count and Flesch Reading Ease Formula) for Navy Enlisted Personnel." Research Branch Report 8-75. Millington, Tennessee: U.S. Naval Air Station, 1975.

Kohn, D. L., and B. S. Sack. "Central Bank Talk: Does it Matter and Why?" Macroeconomics, Monetary Policy, and Financial Stability, Ottawa: Bank of Canada. 2004, 175-206.

Laband, D. N., and C. N. Taylor. "The Impact of Bad Writing in Economics." Economic Inquiry, 30(4), 1992, 673-688.

Morris, S., and H. S. Shin. "Social Value of Public Information." American Economic Review, 92(5), 2002, 1521-1534.

-- "Central Bank Transparency and the Signal Value of Prices." Brookings Papers on Economic Activity, 2, 2005, 1-43.

-- "Coordinating Expectations in Monetary Policy," in Central Banks as Economic Institutions, edited by J.-P. Touffut. Cheltenham, United Kingdom and Northampton, Massachusetts, United States: Edward Elgar, 2008.

Paasche-Orlow, M. K., H. A. Taylor, and F. L. Brancati. "Readability Standards for Informed-Consent Forms as Compared with Actual Readability." The New England Journal of Medicine, 348(8), 2003, 721-726.

Resche, C. "Investigating 'Greenspanese': From Hedging to 'Fuzzy Transparency'." Discourse and Society, 15(6), 2004, 723-744.

Smith Harvey, V. "Improving Readability of Psychological Reports." Professional Psychology: Research and Practice, 28(3), 1997, 271-274.

Stein, J. C. "Cheap Talk and the Fed: A Theory of Imprecise Policy Announcements." American Economic Review, 79(1), 1989, 32-42.

Svensson, L. E. O. "Social Value of Public Information: Morris and Shin (2002) Is Actually Pro Transparency, not Con." American Economic Review, 96(1), 2006, 448-452.

Winkler, B. "Which Kind of Transparency? On the Need for Clarity in Monetary Policy-making." IFO Studien, 48(3), 2002, 401-427.

Woodford, M. "Central Bank Communication and Policy Effectiveness," in The Greenspan Era: Lessons for the Future, Kansas City: Federal Reserve Bank of Kansas City, 2006.


*I thank Jan Mare Berk, Michael Ehrmann, Gabriele Galati, Refet Giirkaynak, Ellen Meade, Carlo Rosa, seminar participants at de Nederlandsche Bank, the University of Amsterdam, and participants in the 40th MMF Group conference for useful comments. This article was greatly improved by comments by three referees and Brad Humphreys (the editor for this article). I also thank the Monetary Policy Research Division of the European Central Bank for its hospitality. Any errors and omissions are my responsibility. Views expressed in this article do not necessarily coincide with those of de Nederlandsche Bank, the European Central Bank, or the Eurosystem.

Jansen: De Nederlandsche Bank, Economics and Research Division, P.O. Box 98, 1000 AB Amsterdam, The Netherlands. Phone 31-20-5243170, Fax 31-20-5242506, E-mail
COPYRIGHT 2011 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2011 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Jansen, David-Jan
Publication:Contemporary Economic Policy
Article Type:Report
Geographic Code:1USA
Date:Oct 1, 2011
Previous Article:The asymmetric impact of "informative" and "uninformative" federal open market committee statements on asset prices.
Next Article:Modeling internal decision making process: an explanation of conflicting empirical results on behavior of non-profit and for-profit hospitals.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters