Journal of Financial Stability 22 (2016) 4556

Contents lists available at ScienceDirect

Journal of Financial Stability

journal homepage: www.elsevier.com/locate/jfstabil

Central bank transparency and nancial stability

Roman Horvth , Dan Vasko
Institute of Economic Studies, Charles University, Prague, Czech Republic

a r t i c l e

i n f o

Article history:
Received 8 October 2014
Received in revised form 6 May 2015
Accepted 10 December 2015
Available online 21 December 2015
JEL classication:
E52
E58
E61

a b s t r a c t
We develop a comprehensive index of the transparency of central banks regarding their policy framework
to safeguard nancial stability for 110 countries from 2000 to 2011 and examine the determinants and
effects of this transparency. We nd that the degree of transparency increased in the 2000s, though
it still varied greatly across the countries in our study. Our regression results suggest that the central
banks that have a transparent monetary policy are more likely to show increased transparency in their
framework for nancial stability. More developed countries exhibit greater transparency, past episodes
of high nancial instability have a negative effect on transparency and the legal origin matters, too. In line
with theoretical literature, our results also suggest a non-linear effect of central bank nancial stability
transparency on nancial stability. If transparency is too high, it is not benecial for nancial stability.
2015 Elsevier B.V. All rights reserved.

Keywords:
Financial stability
Transparency
Central banks

1. Introduction
Central banks around the world have increased the transparency
of their monetary policies in the last two decades. Central banks
nowadays provide very detailed information about various aspects
of their policies. Based on the monetary policy transparency index
for 110 countries from 1998 to 2010, Dincer and Eichengreen (2014)
nd that there is almost no occasion that causes central banks to

We thank two anonymous referees, Alex Cukierman, Adam Gersl, Iftehkar

Hasan, Christian Hott, Ivan Huljak, Petr Jakubk, Branimir Jovanovic, Martin Kuncl,
Diego Moccero, Zeno Rotondi, Anna Samarina, Jakub Seidler, Urszula Szczerbowicz and seminar participants at European Economic Association Annual Conference
(Goteborg, Sweden), The Stability of the European Financial System and the Real
Economy in the Shadow of the Crisis (Dresden), XXI International Conference on
Money, Banking and Finance (Rome), Theories and Methods in Macroeconomics
(Lyon), Czech Economic Society conference (Prague), 6th FIW-Research Conference
(Vienna), European Public Choice Society Annual Conference (Zurich), International Conference on the Global Financial Crisis: European Financial Markets and
Institutions (Southampton), Challenges for Europe conference (Split), EBES conference (Warsaw), 8th Conference on Currency, Banking and International Finance
(Bratislava) and IOS (Regensburg) for helpful comments. Simona Tthov, Josef
Bocek and Nikoloz Kudashvili provided excellent research assistance. We acknowledge the support from the Grant Agency of the Czech Republic P402/12/G097. The
nancial stability transparency index is available from the authors upon request or
directly downloadable from http:// ies. fsv. cuni. cz/ en/ staff/ horvath in the form of
online Appendix.
Corresponding author at: Institute of Economic Studies, Charles University, Opletalova 26, Prague 1 110 00, Czech Republic. Tel.: +420 222 112 317.
E-mail address: roman.horvath@gmail.com (R. Horvth).
http://dx.doi.org/10.1016/j.jfs.2015.12.003
1572-3089/ 2015 Elsevier B.V. All rights reserved.

decrease the degree of their monetary policy transparency. Nevertheless, many central banks are nowadays responsible not only
for monetary policy but more and more frequently for safeguarding nancial stability, too. While the transparency in the area of
monetary policy has been extensively examined, we know much
less about why central banks change their transparency in nancial
stability policies and what are the consequences of these changes.
This is despite the fact that the communication on nancial stability gained the importance dramatically during the crisis (Born et al.,
2014).
In this paper, we construct a central banks transparency index
regarding their framework to safeguard nancial stability for 110
countries from 2000 to 2011. We nd that central banks have
been continuously increasing their transparency in their nancial stability frameworks. Reversals in the trend towards greater
transparency have been extremely rare. The fact that many central
banks have been continuously increasing their transparency does
not invalidate the theoretical literature emphasizing the limits of
transparency (see, among others, Morris and Shin, 2002; Angeletos
and Pavan, 2007; Walsh, 2007; Cornand and Heinemann, 2008;
Cukierman, 2009; van der Cruijsen et al., 2010; Dale et al., 2011)
and the effects of increased transparency on welfare have to be
evaluated also empirically (see Gerlach-Kristen, 2004; Meade and
Stasavage, 2008; Crowe, 2010; Ehrmann et al., 2012 or Horvth
et al., 2012). Unlike in the case of monetary policy transparency, it
is important to comprehensively address the fact that the central
banks are globally not the only institution responsible to safeguard
nancial stability and in some countries they are legally not obliged

46

R. Horvth, D. Vasko / Journal of Financial Stability 22 (2016) 4556

to safeguard nancial stability at all. Therefore, our empirical analysis controls for these sample selection issues.
To contribute to this transparency literature, we examine what
has caused central banks to increase the transparency of their
frameworks to promote nancial stability (for simplicity, we call
this type of transparency nancial stability transparency). Next, we
address whether greater nancial stability transparency is, in fact,
benecial. Although several studies have investigated the determinants and effects of monetary policy transparency (Dincer and
Eichengreen, 2014, among others), this evidence is missing for
nancial stability transparency. Oosterloo et al. (2007) and Cihak
et al. (2012) focus on one particular aspect of nancial stability
transparency, the publication of nancial stability reports (FSRs).
In addition, Cihak et al. (2012) provide a framework to evaluate the
quality of nancial stability reports, but the evidence of the effect
of quality ratings offers only limited support that nancial stability
transparency is benecial. In contrast to these studies, we develop
a comprehensive nancial stability transparency index that focuses
not only on the coverage of nancial stability reports but also on
other communication channels, decision-making procedures and
underlying legal aspects.
In addition, we focus on the effect of central bank transparency
in a more complete manner. To our knowledge, previous literature examined only the effects of monetary policy transparency.
We explore the interactions between the transparency in monetary
policy and nancial stability, as there are several plausible reasons
why transparency in these two areas is likely to be related. Monetary policy transparency increased substantially during the last two
decades, and this trend in monetary policy transparency preceded
that of nancial stability. The rst central banks were assigned the
role of safeguarding nancial stability in the late 1990s or later. If
the central bank makes a decision to communicate more openly
about its monetary policy, this may create an impetus to increase
transparency in other areas of central bank activities, such as in
safeguarding nancial stability. The underlying reason is the ambition of central banks to communicate consistently (Blinder et al.,
2009).
Our results suggest that most central banks have continuously become more transparent in their communication on
issues related to nancial stability. In general, more developed
countries, especially those with Nordic or German legal origin, tend
to exhibit more transparent communications regarding nancial
stability.
The degree of nancial stability transparency depends strongly
on the previous experiences with transparent communications
regarding the banks monetary policy. This result survives a series
of robustness checks such as addressing sample selection issues,
different lag structure or different regression specications. Our
results hold even if we run the regresssions for the components of
nancial stability transparency index thus reducing substantially
the potential concerns that the weighting of index components
matter.
We also nd that there is a non-linear effect of transparency on
the degree of nancial stability, which is in line with the theoretical
literature emphasizing the optimal level of transparency. According to our results, too much transparency may be harmful. This
result is likely to be a consequence of our sample, which includes
the period of global nancial crisis and more transparency about
nancial imbalances and accompanying risks in bad times may, in
fact, escalate the crisis.
This paper is organized as follows. Section 2 discusses reasons
why central banks publish nancial stability reports, develops the
nancial stability transparency indices and presents the resulting
indices. Section 3 presents the regression results on the determinants of nancial stability transparency. Section 4 gives empirical
evidence on whether greater nancial stability transparency

Fig. 1. The number of central banks publishing the nancial stability report.

reduces the risks of nancial instability. Section 5 concludes the

paper. An appendix with additional regression results follows.
2. Central bank transparency index on nancial stability
First, this section focuses on the nancial stability reports published by central banks, as they are an integral part of nancial
stability communications and they strongly inuence the degree
of central bank transparency on nancial stability issues. Second,
this section provides the details on the construction of our nancial stability transparency index. Third, the resulting indices are
presented and discussed.
2.1. Financial stability reports: why publish them?
Many central banks around the world publish nancial stability
reports (FSRs) and use them as their main communication channel
regarding their nancial stability. The Bank of England, for example, characterizes the purpose of its FSR as a means to identify the
major risks to the UK nancial system and to thereby help nancial rms, authorities and the wider public manage and prepare
for these risks. The Swedish Riksbank states that ... FSRs present
the overall assessment of risks and threats to the nancial system and
an evaluation of the capacity for coping with them [...] By making the
analysis available to nancial market participants and other interested
parties, we can share our viewpoints and contribute to the debate
on this subject. Accordingly, Born et al. (2012) suggest that one
important reason why FSRs are published is to effectively guide
the markets and reduce noise.
The rst countries to publish FSRs were the UK and the Scandinavian countries, specically Sweden and Norway. In 1997, Sweden
became the rst country to publish a separate document about its
nancial stability, later called the FSR. Andersson. (2008) contends
that the main reason countries began to publish their FSRs was
related to the nancial crisis of the early 1990s.
Fig. 1 presents the number of countries publishing their FSRs
between 1996 and 2011. It further shows that the rst FSRs
appeared in the late 1990s. The publication of FSRs by central banks
became more common in the 2000s, and currently, more than 60
countries publish an FSR, the vast majority of which are published
by developed countries. Except Ireland, Greece and the USA, all
central banks in OECD countries currently publish their FSRs. The
unavailability of the FSRs from the USA is related to the institutional
setting of nancial sector supervision.1 With respect to Ireland and

1
The FED is not responsible for nancial stability. This role is delegated to the
Financial Stability Oversight Council (FSOC) operating under the Treasury. The FSOC

R. Horvth, D. Vasko / Journal of Financial Stability 22 (2016) 4556

Fig. 2. The number of central banks publishing stress tests and nancial stability
indicators.

Greece, the FSRs were, at one time, available, but central banks
stopped publishing these reports during the current global nancial
crisis.
Cihak (2006) documents that most of the assessments in the
FSRs prior to the recent nancial crisis have been positive (96%
of FSRs assess the nancial sector as in good shape, in solid
shape, or improving). Born et al. (2012) examine the optimism
of FSRs and nd that the degree of optimism was rising during
the 2000s, until the outbreak of the nancial crisis. One way to
address whether this optimism is justied is to conduct nancial
sector stress tests with sufciently adverse scenarios. In this regard,
Breuer et al. (2011) and Franta et al. (2014) propose quantitative
methods how to evaluate whether these scenarios are sufciently
adverse.
In this paper, we examine the FSRs published by central banks.2
The FSRs typically begin with an executive summary of the general assessment of nancial stability and potential risks. The FSRs
then continue with the core analytical aspects, which cover an
assessment of macroeconomic environment and risks and contain
information about various types of institutions (such as deposit takers) and markets important for nancial stability. Other FSRs also
publish several policy-oriented research articles on nancial stability. There are some central banks, that cover do not publish research
articles as the part of FSR (e.g., Norway), while others publish only
research articles (e.g., France). The FSRs almost always contain an
analysis of the banking sector. Nonetheless, other nancial sectors,
such as the non-banking nancial sector, real estate, corporations
and households, are also frequently part of the FSRs.
The analytical portion of the FSR typically contains three types
of indicators to assess stability: soundness indicators, stress tests
and market-based indicators. Using these indicators, the report
should cover all main risks (credit risk, contagion risk, interest rate
risk, liquidity risk, exchange rate risk, payment or settlement risk)
present in the nancial markets. The soundness indicators are a set
of macro-prudential indicators that describe the nancial health
by aggregating indicators from individual nancial institutions. As
a general rule, the (sub)set of nancial soundness indicators proposed by the IMF is used. An increasing number of central banks
have also begun to publish stress tests in their FSRs to assess the
stability of the nancial system. Fig. 2 reports the number of central banks that publish stress tests and nancial stability indicators

publishes an annual report in which the content is very similar to that of the FSRs
published by the central banks. Because we focus on central bank communications
regarding nancial stability, we do not further examine the report of the FSOC (see
Cihak et al., 2012, for further discussion of the US experience). Nevertheless, our
econometric framework addresses the sample selection issues related to the fact
that some central banks are not assigned the responsibility of safeguarding nancial
stability.
2
FSRs are sometimes termed Financial Stability Reviews or Financial Market
Reports. The nancial stability reports by international organizations or private
rms are not considered.

47

within their FSRs. The number of these central banks has been
continuously increasing during the 2000s.
Market-based indicators are also covered in the FSRs, as they
provide useful forward-looking information regarding potential
risks. This group of indicators is typically comprised of the stock
market prices of nancial institutions, the volatility in share prices,
the distance to default, the probability of default and/or distance to
insolvency, various ratings, bond prices and option prices.
One way to evaluate the quality of FSRs is to focus on clarity, consistency and thoroughness of these reports (Cihak, 2006 or Cihak
et al., 2012). In this paper, we mainly focus on the coverage of the
FSRs because it requires less subjective assessments and is more
tractable, thus allowing us to evaluate the FSRs for a wide set of
countries.
2.2. Financial stability transparency index construction
This sub-section presents the construction of our index and its
main descriptive statistics. The source of data are central bank websites. In general, we try to mimic the established methodology for
the construction of monetary policy transparency indexes Eijfnger
and Geraats, 2006 and for the construction of transparency of banking supervisors (Liedorp et al., 2013). The indexes are typically
divided into ve categories: economic, political, procedural, operational and policy transparency. Some of our variables underlying
the aggregate nancial stability transparency index can be also
divided into these categories. For example, the goal of nancial
stability in central bank act represents the political transparency.
We propose the newly constructed nancial stability transparency (FST) index to be the sum of the following 11 items (the
number of points granted for each category is indicated in parentheses):
1. The goal of nancial stability is explicitly stated in the central
bank act (0 not stated, 1 explicitly stated).
2. The publication of the FSR (0 not published; 1 published).
3. The publication of the FSR frequency (0 not published; 1/2
published annually; 1 published semi-annually or more
often).
4. The FSR is forward looking (0 not forward looking, 1 includes
outlooks and forecasts of risks).
5. The coverage of the FSR (in total, a max. of 1.5 points).
(a) Macroeconomic environment and its risks (1/2 if included).
(b) Deposit takers information and its risks (1/2 if included).
(c) Other subjects or market information and risks (1/2 if
included).
6. The publication of stress test (0 stress test not published, 1/2
published annually, 1 published more often).
7. The publication of FSIs (0 not published, 1/2 core set of FSIs
published, 1 both core and encouraged set published).
8. Macro-prudential policy transparency (0 not described, 1/2
general strategy and co-operation described, 1 detailed policy
and crisis management described).
9. The existence of a nancial stability policy committee (0 no
committee, 1 committee with regular meetings and clear
strategy).
10. A separate section on nancial stability on the central banks
website (0 no separate page (section) on web, 1 separate
page on the web).
11. A separate section (database) on the central banks website for
speeches about nancial stability (0 no separate section, 1/2
separate section)
With respect to the goal of nancial stability in the central bank
act (item 1), a country is given one point if the central bank act
states that the central bank is responsible for promoting nancial

48

R. Horvth, D. Vasko / Journal of Financial Stability 22 (2016) 4556

stability. Interestingly, there are many central banks that publish

nancial stability reports, but the act does not stipulate that the
central bank should contribute to nancial stability. As of 2011,
48 countries out of 110 mention nancial stability in the central
bank act. Interestingly, 34 central banks do not mention nancial
stability in the central bank act but do not receive zero points for
all other variables underlying the FST index, so they care about
nancial stability, to a certain extent, even though they are not
obliged by law. In our regression analysis, we examine whether our
results are robust if we exclude these countries and if we formally
address sample selection issues using the Heckman estimator.3
An important component of our index is the periodicity and
coverage of the FSRs (items 2 through 5 of our index), which can
earn 4.5 out of 11 index points, the maximum value of the index.
While only a minority of central banks published their FSRs at the
beginning of our sample (e.g., in 2000), the number of central banks
publishing this report increased to more than one-half of our sample in 2011 (see also Fig. 1). If the FSR is published, it is typically
published once or twice a year. A few central banks publish the
report more frequently, however. In the case of more frequent FSRs,
we decided to assign one point for the value of the index and for the
semi-annual frequency for two reasons. First, we do not believe that
more frequent FSRs increase the transparency substantially, and
second, these reports are usually more concise. In total, 64 countries
published FSRs in 2011, out of which 31 countries published FSRs
annually, and 33 countries published FSRs semi-annually or more
often.
With respect to the forward-looking feature of the report (item
3), we assigned a value of one to FSRs featuring forecasts about
the most important risks to nancial stability. The coverage of FSRs
(item 4) is evaluated according to whether all important segments
of the nancial sector are covered. One-half of a point is assigned
for each separate chapter in the FSRs discussing the macroeconomic
environment and risks, the information and risks related to deposit
takers and other subject information and risks (such as households
and corporations). The average score in 2011 for the content of the
report was 0.72 out of 1.5 points.
The publication of macro stress tests is an important feature of
transparency (item 6), as it gives a quantitative assessment of the
ability of nancial sector to withstand large negative shocks. The
value of 0 is given, if the stress tests are not published. One half is for
the stress tests published annually and the value of one is assigned,
when stress tests are published more often than annual. Our results
show that 33 countries scored one point in 2011 for this item and
more than 40 central banks publish stress tests (see Fig. 2).
We consider the publication of the nancial soundness indicators (FSIs) important because it offers a unique standardized
measure of the current conditions of nancial institutions (item
7).4 The IMF classies the FSIs into two categories: (1) core set and
(2) encouraged set. We assign a value of one-half to those central
banks that published the core set and a value of one to those that
published both the core and the encouraged set. Some central banks
publish these indicators on their website but do not include them

3
On the one hand, the value of 0 in the FST index suggests that the central bank is
not responsible for nancial stability. On the other hand, the ambition of all central
banks is to contribute to price and economic stability. As both price and economic
stability are inuenced by nancial stability, it can be argued that central banks
are at least indirectly concerned about nancial stability. This view is supported by
Kevin Warsh, who stated in his speech delivered at the New York State Economic
Association Annual Conference on October 5, 2007, that It is worth emphasizing that
the Federal Reserves concern with nancial stability stems largely from the adverse
implications of nancial instability for overall economic performance. The Feds interest
in promoting nancial stability is thus intimately connected with its macroeconomic
objectives: maximum sustainable employment and price stability.
4
The study by Oosterloo et al. (2007) use the FSIs as the indicator of the quality
of the FSRs.

(or include only selected indicators) in the FSR. Only 12 countries

published the core set in 2011, and half of them also published the
encouraged set on the central bank website.
Transparency regarding macroprudential policy is an additional
element of our index (item 8). If the macroprudential policy framework is described on the central banks website, the country earns
0.5 points for the index. To obtain one point, the policy must contain information about crisis management with the precise roles
of all participating institutions being identied. The score of the
countries for this item is rather low, with an average of 0.17 in
2011.
A stand-alone nancial stability committee with regular meetings is another variable underlying our FST index (item 9). Central
banks that establish a nancial stability committee signal transparently to the markets who is responsible for making the decisions
related to promoting nancial stability. However, only the UK,
Ireland and Portugal scored a non-zero value for this item by having
a separate committee with regular meetings. It is important to note
that the composition of a nancial stability committee is not necessarily identical to the monetary policy committee. For example, the
nancial stability committee of the Bank of England includes several senior managers who are not members of the monetary policy
committee.
Next, a value of one is assigned to those central banks that use
their website effectively for communication regarding their nancial stability (item 10). We operationalize this issue by examining
whether central banks have a separate webpage (or webpages)
solely dedicated to nancial stability. This page(s) should contain all
important information on nancial stability, such as the denition
of nancial stability and its importance for economic development
or FSRs. Interestingly, despite Born et al. (2014) document the
important role the central bank communication plays for nancial stability only 38 countries have a separate section on nancial
stability on their webpages.
Finally, we assign a half point for those central banks that have
a database of speeches (item 11) made by their central bank representatives divided according to topic, including nancial stability.
Only 5 countries received a half point for the speeches database.
Fig. 3 summarizes the construction of our FST index.
We are aware that there are many issues when constructing the
indices. The well-known issue is the choice and normalization of
underlying indicators that form the aggregate index. With respect
to normalization, we opted for simple averages because of their
transparency. In addition, we want to produce an index that is comprehensive but not extremely difcult to replicate and update. We
are aware that there may be some cases when the index does not
fully address the changes in transparency. For example, the central
bank may be less transparent regarding the analytical background
behind the stress tests or the stringency of the stress tests. One
way to put aside the concerns about the choice and normalization
of underlying indicators is to explain the nancial stability transparency index and its components using regression analysis. If we
cannot nd plausible determinants of the FST index, then the critics
may be right.
For the selection of countries, we follow Dincer and Eichengreen
(2014), who develop the monetary policy transparency index based
on the methodology of Eijfnger and Geraats (2006). We choose
an identical set of 110 countries with the yearly frequency of the
index (their 100 countries and additionally the euro area countries,
European Central Bank monetary policy transparency is used for
the euro area countries). This allows us to evaluate our hypothesis on the transparency culture in central banks. More specically,
one of our hypotheses (more on this in the following section) is
that central banks that are already comfortable with being transparent are more likely to become transparent in the new areas of
their business, such as safeguarding nancial stability. To examine

R. Horvth, D. Vasko / Journal of Financial Stability 22 (2016) 4556

49

Fig. 3. Financial stability transparency index: an overview.

Fig. 4. The FST index. OECD vs. non-OECD countries. EU and ination targeters.

this hypothesis, we use the monetary policy transparency index to

proxy for the prevailing transparency culture in the central banks
(the monetary policy transparency became an issue in the central
banks in the 1980s to 1990s, i.e., well before the considerations
about transparency on nancial stability issues).
2.3. Financial stability transparency index results
The detailed country-level results for the FST index are available in the online Appendix. Some gures summarizing the main
results are available below. Fig. 4 shows that, according to our index,
central bank transparency regarding their policy framework to promote nancial stability has been continuously increasing over time.
The transparency was rather low in 2000, with an average score
for the FST index of approximately 2 out of 11 points. In contrast
to monetary policy transparency, the communication on nancial
stability is a new phenomenon that has risen in importance during
the last decade. We nd that the EU, OECD and ination-targeting
countries are the most transparent and that the transparency gap
between them and non-OECD countries has slightly increased over
time.
Fig. 5 presents a list of top performers. Not surprisingly, developed countries score very high on our index. The UK is a top
performer, with 9.5 out of 11 points, followed by several Central
European and Nordic countries. Interestingly, many Central and
Eastern European countries appear in the list of top performers.
This is not surprising, however, as this group of countries currently
exhibits a very high degree of monetary policy transparency as well
(Siklos, 2011). The U.K. and Nordic countries score very high in
terms of monetary policy transparency (Dincer and Eichengreen,
2014) and, according to Liedorp et al. (2013), who provide an index
of banking supervision transparency for 24 countries, also have
highly transparent bank supervision.

Next, we compare the FST index with respect to legal origin,

dened as in Dincer and Eichengreen (2014). It seems that the
countries with Nordic legal origin, followed by those with German
legal origin, typically exhibit high values on the index of between
4 and 5. On the other hand, the countries with French, English and
Socialist origin display values between 2 and 2.5.
As argued above, one of our hypotheses is that the central banks
that are transparent in their other activities, such as in monetary
policy conduct, are more likely to be transparent about nancial
stability issues. In our opinion, there are two underlying factors
consistency in communication and culture. The ambition of the
central bank is to communicate consistently. It would be difcult
for central banks to explain to the public why their communication
is transparent in some areas but not in others. The central bank
culture might also play a role. In some countries, there is an established culture of accountability of central bankers that is closely
related to transparency, and therefore, central banks might be
accustomed to communicating and acting transparently. It is well
known that central banks increased their monetary policy transparency in the 1980s and 1990s, i.e., well before the discussions
about nancial stability transparency appeared. For this reason, we
use the monetary policy transparency index to proxy for the effects
of central bank transparency culture as well as for the consistency
in communication, and we examine the effect of monetary policy
transparency on nancial stability transparency.
As previously mentioned, we use the monetary policy transparency index (MPT index) by Dincer and Eichengreen (2014). The
contemporaneous correlation of the MPT index and the FST index
is 0.59, which is signicantly different from zero at the 1% level (see
also Fig. 6). The correlation is 0.64 if we restrict the sample to the
central banks, which are obliged to safeguard nancial stability by
law. The correlation of the FST index with the MPT index at various
lags reaches similar values.
3. What drives central bank transparency on nancial
stability issues?
This section investigates the determinants of central bank
transparency on nancial stability issues. Oosterloo et al. (2007)
examine those factors that contribute to the publishing of the
FSR and nd that the experience of a past banking crisis, higher
income per capita and EU membership increase the likelihood
of the FSR being published. We also examine the determinants
of why FSRs are published. In some sense, whether the FSRs are
published may be considered as a crude indicator of nancial
stability transparency; thus, it provides a robustness check to our
baseline estimations for which we use our FST index as a dependent

50

R. Horvth, D. Vasko / Journal of Financial Stability 22 (2016) 4556

6
4
0

FST index

10

Fig. 5. Countries with a highest nancial stability transparency index. Note: Index values as of 2011.

10

15

MPT index

Fig. 6. Financial stability transparency vs. monetary policy transparency. Note:

Financial stability transparency is proxied by the index developed in this paper.
Monetary policy transparency is represented by the index developed by Dincer and
Eichengreen (2014).

variable. As a consequence, this extends the research by Oosterloo

et al. (2007) on understanding nancial stability transparency.
In addition, we also use a broader set of regressors to analyze
the motive for greater transparency. Our supposition is that central bank culture matters. In particular, we analyze the effect of
monetary policy transparency, assuming that a well-established
and transparent approach towards the communication of monetary policy is likely to lead to more transparent communication on
nancial stability. Monetary policy transparency preceded nancial stability transparency, as many countries around the world had
already substantially increased their monetary policy transparency
in the 1990s (Blinder et al., 2009; Crowe and Meade, 2008).
We expect that transparency may be inuenced by previous
experience with nancial instability. Accordingly, the central banks
may be reluctant to increase transparency when nancial markets
are under severe distress, which could escalate the crisis further.
We use the World Bank data on banking sector crisis (dened as
the dummy variable) and IMF nancial stress index developed by
Balakrishnan et al. (2009) and Cardarelli et al. (2011) because the
index is comprehensive and, compared to other nancial stress
indices, has a solid country coverage.5

The index is available for 17 advanced and 26 emerging countries until 2009. The
index comprises the following seven components: banking sector beta (standard
beta from capital asset pricing model), TED spread (3-month LIBOR minus the
government short term rate), inverted term spread (government short term rate

Next, as a proxy of the importance of the nancial sector, we use

the ratio of stock market capitalization to GDP and credit to GDP
ratio published by the World Bank. Our hypothesis is that more
transparent communication towards nancial markets increases in
importance in nancially developed economies. The GDP per capita
is included to proxy for the level of economic development. In addition, we control for the institutional structure of nancial market
supervision using the World Bank data (see Melecky and Podpiera,
2013). This is important because central banks may be more transparent in their framework to promote nancial stability, if they
have information and tools to combat nancial crisis (Cukierman,
2009). For this reason, we construct a variable capturing to what
extent central bank is involved in the nancial markets supervision.
We assign the value of one, if nancial market supervision is fully
under the umbrella of central bank. The value of 0.5 is assigned, if
central bank supervises only banks. We assign the value of 0, if central bank is not involved in supervision. Next, we also control for
macroeconomic stability by including the GDP growth and ination and for the degree of openness of the economy (the source for
these data is World Bank database).
We estimate the following baseline model:
FSTindexi,t = Xi,tj + i,t + ei,t

(1)

FSTindexi,t denotes our nancial stability transparency index for

country i in time (year) t. The explanatory variables are: Xi,tj . For
our baseline model, we assume j = 0. A robustness check with j = 3
is conducted. As previously discussed, our list of explanatory variables includes a monetary policy transparency index, banking crisis
dummy, GDP per capita, supervisory structure, stock market capitalization to GDP, credit to GDP ratio, openness, ination and GDP
growth. We use the xed effects estimator to estimate Eq. (1).
The regression results on the determinants of the nancial stability transparency index are available in Table 1. We nd that
monetary policy transparency contributes positively to nancial
stability transparency. Further, the results suggest that central
banks in developed countries exhibit higher transparency in their
policy framework on nancial stability. The experience of banking sector crisis in the past is associated with lower transparency
scores. We nd some evidence that central banks, which are more
involved in the nancial market supervision, are more likely to
display higher transparency scores. These central banks typically

minus government long-term rate), corporate debt spreads (corporate bond yields
minus long-term government bond yields), stock market change (month-overmonth change multiplied by minus one), stock market volatility (6-month moving
average of the squared month-on-month returns) and exchange market volatility (6month moving average of the squared month-on-month growth rate of the exchange
rate). A higher value for the nancial stress index implies a higher degree of stress.

R. Horvth, D. Vasko / Journal of Financial Stability 22 (2016) 4556

51

Table 1
What determines nancial stability transparency? Note: The dependent variable is the nancial stability transparency index. A restricted sample denotes a sample of central
banks with, which are obliged by law to safeguard nancial stability. t-statistics are shown in parentheses; ***, ** and * denote statistical signicance at the 1%, 5% and 10%
levels, respectively. Fixed effects estimation. Explanatory variables lagged by 0 or 3 years.
FST index

(I)

(II)

(III)

(IV)

Full sample

MP transparency
GDP p.c.
Banking crisis
Supervisory structure

(VI)

(VII)

Lagged by 3 years

Contemporaneous

Lagged by 3 years

0.63***
(14.33)
0.08***
(9.77)
0.11
(0.63)
1.73
(1.58)

0.54***
(11.14)
0.02***
(2.79)
0.07
(0.32)
0.79
(1.61)

0.76***
(10.86)
0.06***
(4.47)
0.02
(0.96)
0.58
(0.95)

0.64***
(9.52)
0.01
(0.20)
0.67**
(2.39)
0.55
(0.91)

Openness
Credit to GDP ratio
Ination
GDP growth
884
0.22

(VIII)

Restricted sample

Contemporaneous

Market capitalization

No. of observations
R-sqr. adj.

(V)

0.66***
(11.29)
0.07***
(5.95)
0.24
(1.15)
1.18**
(2.25)
0.01
(0.66)
0.01
(0.14)
0.01***
(3.09)
0.01
(0.43)
0.01
(1.50)
665
0.17

693
0.19

have more tools and possibly better information on how to combat

the nancial imbalances. Credit to GDP ratio matters for transparency, to a certain extent. When we exchange the banking crisis
dummy for IMFs nancial stress index in our regression, we nd
some limited evidence that the episodes of high nancial stress
negatively affect nancial stability transparency (the results are
available upon request).
We estimate our regression model both for the full sample and
the restricted sample. The central banks, which are not obliged to
safeguard nancial stability, are excluded from the restricted sample. It is important to also run the regressions for the restricted
sample because some central banks are not responsible for nancial
stability by law, and therefore, they exhibit low nancial stability
transparency. On the other hand, there is a rationale for estimating
our regressions based on the full model because central banks are
at least indirectly concerned about nancial stability, as it has an
effect on price stability and economic activity. To a large degree,
our ndings regarding the signicance of monetary policy transparency, credit to GDP and economic development do not depend
on whether we use full or restricted sample. On the other hand,
the effect of institutional structure of nancial markets supervision
holds only for the full sample of central banks (more precisely, in
two cases the corresponding p-value is greater than 0.10 but only
mildly). This result suggests that nancial stability transparency
increases in relative terms especially for those central banks, which
supervise the nancial sector (i.e. are involved in microprudential
supervision) but, according to the central bank law, are not obliged
to safeguard nancial sector as a whole. Therefore, we observe a
substitution effect. The central banks become transparent either
if they are required by law to promote nancial stability or if they
are required to supervise the nancial sector. Relatedly, the banking crisis does not have an effect on nancial stability transparency
globally but central banks are more reluctant to increase their
transparency once they are charged legally to safeguard nancial
stability.
To address the selection issues further, we run a two-step Heckman estimation to address the fact that the central bank nancial
stability transparency depends on whether nancial stability is
included as a goal in the central bank act. The governments decision
on whether nancial stability is included in the central bank act is

0.49***
(8.01)
0.01
(0.31)
0.02
(0.08)
1.19**
(2.27)
0.01
(0.87)
0.01
(0.92)
0.01*
(1.88)
0.01
(0.63)
0.01**
(2.05)
538
0.10

342
0.30

0.74***
(8.05)
0.05***
(2.81)
0.66**
(2.10)
0.79
(1.23)
0.01
(0.39)
0.01
(0.74)
0.01
(0.92)
0.01
(0.62)
0.01
(0.24)
271
0.22

308
0.25

0.49***
(6.69)
0.01
(1.01)
0.70**
(2.14)
0.52
(0.94)
0.01
(0.87)
0.01
(0.85)
0.01**
(2.24)
0.01
(0.51)
0.01
(0.04)
247
0.17

likely to be non-random and can be inuenced by specic events,

such as the occurrence of past nancial stress, the experience of
deep recessions caused by nancial instability or nancial depth.
The legal origin may also inuence government actions regarding
nancial stability. Therefore, our selection equation explains the
dummy variable, which takes the value of one if nancial stability is stated as a goal in the central bank act. Our second-stage
regressions posses largely the same set of explanatory variables
as those in the baseline regressions. The results are presented in
Table 2 and, to a large extent, conrm the baseline results that
monetary policy transparency is positively related to nancial stability transparency. On the other hand, other regressors lose their
signicance.
Next, we run the cross-sectional regression to examine the effect
of legal origin on nancial stability transparency given that legal
origin is time-invariant. The results are presented in Table 3. To
address the endogeneity of the regressors other than legal origin,
the dependent variable is averaged over 20012011 and explanatory variables are as of 2000. Our results suggest that Nordic and
German legal systems tend to be positively related to nancial
stability transparency (even though the signicance varies somewhat), while for English and French legal systems, we observe a
negative or zero effect. There are many former colonies of English
and French legal origin in our sample, which is likely a factor in the
insignicance of these two variables. The results on legal origin are
somewhat in contrast with that of Oosterlo et al. (2007), who do
not nd any effect of legal origin on the decision to publish nancial stability reports. The effects of legal origin are estimated using
cross-sectional regressions specied in the way that regressors do
not suffer from any potential endogeneity so that the results can be
interpreted causally.
We carry out a number of robustness checks. First, because our
nancial stability transparency index may be too nely measured,
we estimate the factors inuencing the publication of the FSR, see
Table 5 in the Appendix. If the FSR is published in year t, we assign
a value of one; otherwise, we assign a value of zero. Therefore, we
use a logit model to address the limitations of the dependent variable. As a consequence, this regression model closely resembles
that of Oosterloo et al. (2007), though we use a wider set of determinants, as we, for example, include a monetary policy transparency

52

R. Horvth, D. Vasko / Journal of Financial Stability 22 (2016) 4556

Table 2
What determines nancial stability transparency? Heckman Selection model. Note: The dependent variable is the nancial stability transparency index. The dummy, that is,
whether nancial stability is included in the central bank activity, serves as a dependent variable in the selection equation. Explanatory variables are lagged by 0 and 3 years,
respectively. t-statistics are shown in parentheses; ***, ** and * denote statistical signicance at the 1%, 5% and 10% levels, respectively. A two-step method. Pooled data.
FST index

(I)

(II)

(III)

(IV)

Comtemporaneous
Second stage: nancial stability transparency
MP transparency
GDP p.c.
Banking crisis

0.49***
(10.42)
0.04
(0.62)
0.07
(0.18)

Market capitalization
Supervisory structure
First stage: nancial stability in the central bank act
0.20
Banking crisis
(1.25)
0.02***
Market capitalization
(3.00)
GDP p.c.
0.01
(1.35)
Supervisory structure
0.77***
(3.91)
1.35***
Legal origin French
(8.42)
0.40**
Legal origin German
(2.36)
1.42***
Legal origin English
(8.20)
1.82***
Mills ratio
749
No. of observations

Lagged by three years

0.48***
(10.40)
0.01
(0.75)
0.28
(0.67)
0.01
(1.40)
0.65
(1.30)

0.47***
(9.19)
0.01
(0.64)
0.78
(1.43)

0.46***
(9.08)
0.02
(0.27)
1.01*
(1.80)
0.01
(1.43)
0.67
(1.20)

0.20
(1.25)
0.02***
(3.00)
0.01
(1.35)
0.77***
(3.91)
1.35***
(8.42)
0.40**
(2.36)
1.43***
(8.20)
2.05***
749

0.19
(0.87)
0.02**
(2.41)
0.01
(0.67)
0.66***
(3.11)
1.34***
(7.50)
0.41**
(2.13)
1.38***
(7.21)
2.09***
615

0.06
(1.11)
0.02**
(2.41)
0.01
(0.67)
0.66***
(3.11)
1.32***
(7.54)
0.38**
(2.05)
1.35***
(7.22)
2.23***
615

Table 3
What determines nancial stability transparency? The effect of legal origin. Note: The dependent variable is the nancial stability transparency index. A cross-sectional
regression with the dependent variable is averaged over 20012011. Explanatory variables as of 2000. A restricted sample denotes a sample of central banks, which are
obliged to safeguard nancial stability by law. t-statistics are shown in parentheses; ***, ** and * denote statistical signicance at the 1%, 5% and 10% levels, respectively.
FST index

(I)

(II)

(III)

(IV)

Full sample
MP transparency

0.36***
(6.57)

GDP p.c.
Banking crisis
Market capitalization
Supervisory structure
Legal origin French
Legal origin German
Legal origin English
Legal origin Nordic
No. of observations
R-sqr. adj.

0.61
(1.11)
1.82***
(2.65)
0.78
(1.34)
3.27***
(3.89)
91
0.25

1.38**
(2.42)
0.58
(0.92)
1.32**
(2.25)
1.56**
(2.02)
91
0.49

index. Second, we re-estimate our baseline model without a measure for nancial stress (not reported). The nancial stress index
is available only for approximately half of the countries, which, as
a consequence, substantially reduces the number of observations.
Third, we re-estimate our baseline regression model with explanatory variables lagged by three years (see the results in Table 1).
Forth, we run the regressions with the separate components of
FST index as the dependent variable. In line with Fig. 3, we create four components of the index: General framework, nancial
stability reports coverage, tests and indicators, and website. This is

(V)

(VI)

Restricted sample
0.31***
(4.52)
0.01
(0.58)
0.11
(0.15)
0.001
(0.65)
1.44
(1.31)
2.31***
(3.89)
0.46
(0.72)
1.79***
(3.06)
0.54
(0.60)
68
0.46

0.33***
(5.69)

0.15
(0.26)
1.28***
(2.64)
0.01
(0.01)
3.27***
(3.87)
74
0.20

0.81
(1.36)
0.68
(1.06)
0.67
(1.18)
1.71**
(2.14)
74
0.42

0.32***
(3.82)
0.01
(0.61)
0.11
(0.16)
0.001
(0.64)
1.49
(1.30)
1.89***
(3.89)
0.50
(0.79)
1.51**
(2.43)
0.61
(0.67)
61
0.44

an important robustness check because a critique may say that the

weighting of index plays a role for our baseline ndings. The results
are presented in Table 6 in the Appendix.6 The components of FST
index are positively correlated but the correlation is far from zero
(see Table 7 in the Appendix). Again, our regression results largely
support our baseline ndings and our index is likely to be robust to

6
The regressions were not estimated for the component Website. This components does not display sufcient variation.

R. Horvth, D. Vasko / Journal of Financial Stability 22 (2016) 4556

the different weighting scheme of underlying components. Monetary policy transparency and the level of transparency is signicant
in all specications, while others matter only for the selected components. Interestingly, our results indicate that supervisory structure and market capitalization matter only for the subcomponent
Tests and Indicators. Finally, we re-estimate our regressions using
ArellanoBond estimator. The lagged value of FST index is highly
signicant with the coefcient estimate of approximately 0.75. The
monetary policy index remains signicant with the expected sign.
All in all, the battery of robustness checks largely support our
baseline results. The monetary policy transparency index is statistically signicant in all regression specications, and its coefcient
is, to a large extent, stable. More developed countries are found to
exhibit more transparent communication regarding nancial stability in many regressions. Other explanatory variables matter only
in specic regressions.
4. Does nancial stability transparency have an effect on
nancial stability?
While the previous section focused on why some central banks
are more transparent than others, this section examines whether
the transparency in policy frameworks on nancial stability is benecial. Born et al. (2014) show that the optimistic Financial Stability
Reports contribute to positive and sometimes long-lasting effects
on stock markets and thus indirectly have an effect on economic
activity. Claessens et al. (2013) show how individual macroprudential measures contribute to nancial stability.
We estimate panel regressions to assess the effect of our nancial stability transparency index on nancial instability. More
specically, we focus on whether our index helps explain (1) the
share of non-performing loans, (2) nancial stress and (3) banking
crisis, while controlling for some economic and nancial variables.
We introduce the FSTindexi,tj as well as its squared value to evaluate whether there is an optimal level of transparency.
Therefore, we estimate the following model:
Finstressi,t = Xi,tj + i,t + ei,t

(2)

where Finstressi,t represents a measure of nancial instability for

country i in time (year) t: the share of non-performing loans and
the IMF nancial stress index, respectively. j represents the number of lags, and we set j = 0 in the baseline regressions and j = 3 as
a robustness check. Xi,tj is a vector of regressors, which includes
2

FSTindexi,tj and FSTindexi,tj . In line with previous literature (e.g.,

Vazquez et al., 2012), the non-performing loans ratio (NPLt ) is transformed as ln(NPLt /(1 NPLt )). We control for a number of variables:
GDP per capita, market capitalization, openness, credit to GDP ratio,
real interest rate, ination, GDP growth and nominal exchange rate
change, as these variables typically belong to the set of early warning indicators in regressions examining the likelihood of nancial
crisis (Frankel and Saravelos, 2012). In addition, to the estimation in
Eq. (2), we also analyze whether nancial stability transparency is
associated with banking crisis. For this reason, we estimate panel
logit regressions with the dependent variable being the banking
crisis dummy. Except transparency scores, our data are from the
World Bank databases.
We estimate Eq. (2) for both full and restricted samples. We
dene the restricted sample such that we include only those central
banks, which are obliged to promote nancial stability by law.
Our regression results are provided in Table 4. In most specications, greater nancial stability transparency is found to contribute
to lower nancial stress and to fewer bad loans. On the other hand,
the squared index of nancial stability transparency is positive.
This result suggests a non-linear effect of transparency on stability
and that too much transparency can harm stability. The estimated

53

coefcients imply that the optimal level of transparency is between

5 to 8 points depending on the specication. We do not want to
overemphasize this result because the estimates are surrounded
by a certain degree of uncertainty.
In general, the results are in line with the theoretical literature
emphasizing that there is an optimal level of bank transparency
(Cukierman, 2009; van der Cruijsen et al., 2010; Walsh, 2007) and
that under some conditions too much transparency leads to an inefcient liquidation of banks. For our sample covering the global
nancial crisis, our nding of the non-linear effect of nancial
stability transparency may reect that too transparent communication about nancial stability during the period of high nancial
distress may in fact escalate the crisis. For example, the Deutsche
Bundesbank decided not to publish their nancial stability review
in 2008 because of the high nancial stress. Similarly, the nancial stability reports are not available for Ireland during the recent
global nancial crisis. Gick and Pausch (2012) and Goldstein and
Sapra (2014) set up theoretical models to shed light on this phenomenon and show that the public disclosure of macro stress tests
is benecial for market discipline unless there are excessive informational and market frictions.
As a result, the optimal level of transparency is likely to depend
on the business or nancial cycle. The optimal level of transparency
is likely to be higher in good times than in bad times. In bad
times, releasing additional information about nancial imbalances
and accompanying risks may even escalate the crisis. This is in
line with our observation that some central banks in countries hit
by the nancial crisis most strongly decreased their transparency
regarding their framework to support nancial stability. An interesting observation from the policy perspective is that it is far from
easy to decrease the level of transparency. This decision is likely to
be associated with some reputational costs.
To some degree, our empirical exercise is similar to that of Cihak
et al. (2012), who created the nancial stability report composite
quality rating for 44 countries. This quality rating is different from
our nancial stability transparency index, as our index does not
focus as heavily on nancial stability reports and is available for 110
countries. Furthermore, our index is less subjective, as it focuses
more on the coverage of nancial stability reports than on the quality of those reports. On the other hand, the approach presented by
Cihak et al. (2012) is more ambitious in that it focuses on the clarity and quality of the nancial stability reports and requires careful
reading of and expert judgment of all nancial stability reports.
Cihak et al. (2012) examine whether the publication of nancial
stability reports or the nancial stability report composite quality rating inuences the occurrence of banking crises and various
measures of nancial market volatility. They nd that the publication of nancial stability reports does not have an effect on nancial
stability. In addition, their results indicate that the nancial stability report composite quality rating is positively related to nancial
stability to a certain extent. In turn, their results provide some
support to the supposition that clear and consistent central bank
communication regarding nancial stability is benecial.
Compared to the results of Cihak et al. (2012), our results provide
a richer perspective on the effect of nancial stability transparency
because we show that the effects of nancial stability transparency
may differ according to the level of transparency.
In addition, our results also indicate that GDP per capita and
credit to GDP ratio are positively linked to nancial instability. This
result is likely to be related to the fact that the recent global nancial crisis more heavily impacted developed (European) countries.
On the other hand, interestingly, more developed stock markets
(as proxied by market capitalization) are found, in our regressions,
to be benecial to nancial stability, as deeper nancial markets
are more able to absorb shocks. Some other explanatory variables
matter, to a certain extent, too. High ination, high real interest

54

R. Horvth, D. Vasko / Journal of Financial Stability 22 (2016) 4556

Table 4
Does nancial stability transparency have a non-linear effect on nancial instability? Note: t-statistics are shown in parentheses; ***, ** and * denote statistical signicance
at the 1%, 5% and 10% level, respectively. Fixed effects estimation. The dependent variable, non-performing loans, is divided by total loans and logit transformed. Panel logit
regressions for the determinants of banking crisis; pseudo-R-sqr. presented intead of R-sqr. adj.
(I)

(II)

Full sample

(III)

(IV)

Full sample

0.31***
(4.52)
0.03***
(3.13)
0.01
(1.19)
0.01***
(4.49)

0.53***
(2.67)
0.05**
(2.13)
0.11***
(5.61)
0.03***
(7.04)

Non-perform. loans
Fin. stab. transparency
Fin. stab. transparency sqr.
GDP p.c.
Market capitalization

0.19***
(3.91)
0.01**
(2.29)
0.01**
(2.30)
0.01***
(6.88)

Openness
Credit to GDP ratio
Real interest rate
Ination
GDP growth
Exchange rate change
No. of observations
R-sqr. adj.

728
0.37

0.19***
(2.89)
0.02***
(2.36)
0.03***
(3.23)
0.01***
(5.44)
0.01
(1.43)
0.01***
(4.29)
0.02***
(3.67)
0.01
(0.87)
0.01***
(3.23)
0.01
(1.32)
435
0.26

(V)

Restricted sample

(VI)

(VII)

(VIII)

527
0.28

387
0.07

rates and low growth have certain negative effect on our measures
of nancial stability.
According to our results, nancial stability transparency is not
associated to banking crisis. This result suggests that central bank
communication may increase or decrease the degree of nancial
instability, to a certain extent, but neither it helps avoid crisis, nor
it helps trigger crisis.
We carry out a robustness check and lag the regressors by
three period. The nancial stability transparency exerts a nonlinear effect on nancial instability but the effect is sometimes not
statistically signcant at conventional levels. In addition, we also
include explanatory variables capturing the banking sector concentration and loan to deposit ratio. The results on the effect of nancial
stability transparency remain unchanged and these newly added
variables are not found to be systematically signicant. Finally, we
re-estimate the regressions using Arellano-Bond estimator and nd
that nancial stability still exert the non-linear effect on nancial
stability. The results are available upon request.

5. Concluding remarks
In this paper, we develop an index that assesses the degree of the
transparency of central banks regarding their policy frameworks
to safeguard nancial stability for 110 countries in 20002011.
The index consists of several items, such as the coverage of nancial stability reports and the forward-looking orientation of these
reports, the availability of stress tests, the provision of nancial stability indicators to the public, and the clarity of the role of central
banks in safeguarding the nancial stability and decision-making
procedures. We posit that the degree of nancial stability transparency is related to monetary policy transparency. Using these
indices, we rst investigate what determines the nancial stability transparency of central banks. Second, we examine whether
greater nancial stability transparency is benecial to the stability
of nancial markets.

0.06
(0.26)
0.05*
(1.80)
0.11***
(3.18)
0.07***
(7.26)
0.02
(0.62)
0.03***
(2.65)
0.05
(1.18)
0.24***
(4.34)
0.01***
(2.75)
0.01
(1.11)
222
0.13

(X)

Full sample

0.76***
(2.91)
0.08***
(2.54)
0.11***
(4.99)
0.03***
(5.67)

0.05
(0.19)
0.06
(1.59)
0.37***
(6.54)
0.06***
(6.15)

Financial stress index

0.24***
(2.91)
0.03***
(2.77)
0.04***
(4.22)
0.01***
(3.75)
0.01
(0.90)
0.01***
(4.50)
0.06***
(5.36)
0.01
(0.31)
0.01
(1.55)
0.01
(1.53)
317
0.24

(IX)

Restricted sample

(XI)

(XII)

Restricted sample

Banking crisis dummy

312
0.08

0.06
(0.17)
0.06
(1.52)
0.08**
(2.16)
0.04***
(5.99)
0.04
(0.90)
0.03**
(2.37)
0.06
(1.02)
0.21***
(3.13)
0.01
(0.92)
0.01
(1.03)
166
0.14

374
0.71

1.45
(1.39)
0.19
(1.46)
0.28*
(1.66)
0.18***
(3.38)
0.15**
(2.15)
0.16**
(2.84)
0.06
(0.97)
0.13
(1.18)
0.01**
(2.04)
0.01
(0.61)
138
0.14

0.63
(1.56)
0.02
(0.34)
0.35***
(6.02)
0.06***
(5.40)

303
0.43

2.41
(1.14)
0.36
(1.19)
0.64
(1.33)
0.34**
(2.48)
0.30**
(2.27)
0.28**
(2.39)
0.16
(1.48)
0.30*
(1.81)
0.01*
(2.20)
0.01
(0.62)
110
0.79

Our results suggest that nancial stability transparency has been

continuously increasing during the 2000s, though it still varies
greatly across central banks. Furthermore, nancial stability transparency is greater in more developed countries, especially in those
with Nordic and German legal origin. We nd some evidence that
past episodes of nancial instability tend to have a negative effect
on how transparent central banks are regarding their framework
to promote nancial stability. Next, our results show that central
banks with more transparent monetary policies tend to exhibit
greater nancial stability transparency. This result, although new,
is not surprising. It is plausible that central banks, which are used to
communicating transparently in some areas of their activities (e.g.,
their monetary policy), will transmit transparency to new areas of
their businesses (e.g., nancial stability). We also nd that central
banks are more transparent if they are involved in nancial markets
supervision and regulation. Detailed information about nancial
markets and the appropriate tools to address the nancial imbalances are necessary condition for central banks to increase their
transparency. Next, our results indicate that the consequences of
greater nancial stability transparency on nancial stress depend
on the degree of transparency. The effect of transparency is positive
unless the transparency is too high. As a consequence, this nding
is in line with the theoretical literature emphasizing the optimal
level of transparency.
In terms of future research, it would be worthwhile to examine
the transparency of nancial stability policies of other institutions
involved, such as the government, independent supervisory agencies or international organizations.

Appendix
Table 5
Table 6
Table 7

R. Horvth, D. Vasko / Journal of Financial Stability 22 (2016) 4556

55

Table 5
What determines the publication of nancial stability reports? Note: The dependent variable is the nancial stability report dummy. Explanatory variables are lagged by 0
year and 3 years, respectively. A restricted sample denotes a sample of central banks obliged to safeguard nancial stability by law. t-statistics are shown in parentheses; ***,
** and * denote statistical signicance at the 1%, 5% and 10% levels, respectively. Fixed effects logit estimation.
FSR dummy

(I)

(II)

(III)

(IV)

(V)

(VI)

(VII)

Full sample

MP transparency
GDP p.c.
Banking crisis
Supervisory structure
Market capitalization

Contemporaneous

Lagged by three years

Contemporaneous

Lagged by three years

2.19***
(5.44)
0.51***
(5.42)
0.79
(1.10)
2.57
(1.11)
0.01
(0.51)

1.97***
(5.71)
0.07
(1.40)
0.10
(0.13)
5.85***
(2.70)
0.02***
(2.51)

2.99**
(2.34)
1.51***
(3.26)
0.37
(0.27)
4.57
(0.31)
0.02
(1.22)

1.71***
(2.68)
1.85***
(2.79)
9.04
(1.02)
4.93
(1.43)
0.02
(1.61)

1.93***
(4.44)
0.43**
(4.56)
1.05
(1.16)
1.45
(0.57)
0.01
(0.11)
0.01
(0.12)
0.02
(1.20)
0.01
(0.19)
0.02
(1.07)
488
0.60

Openness
Credit to GDP ratio
Ination
GDP growth
No. of observations
Pseudo R-sqr.

(VIII)

Restricted sample

596
0.66

1.61***
(3.88)
0.09
(0.89)
0.10
(0.13)
5.60**
(2.26)
0.02**
(2.09)
0.19***
(3.05)
0.01
(0.20)
0.05
(0.91)
0.01
(0.47)
387
0.78

341
0.45

244
0.82

7.77**
(2.45)
5.32**
(2.55)
3.61
(1.40)
10.47
(0.01)
0.02
(0.98)
0.17
(1.37)
0.17**
(1.99)
0.21
(1.40)
0.01
(1.48)
194
0.87

159
0.56

1.67
(1.47)
4.63**
(2.20)
25.4
(0.05)
18.24**
(2.32)
0.04
(1.29)
0.75**
(2.51)
0.12
(1.32)
0.02
(0.14)
0.01
(0.98)
98
0.76

Table 6
What determines the components of nancial stability transparency index? Note: The dependent variable is the nancial stability transparency index. A restricted sample
denotes a sample of central banks, which are obliged to safeguard nancial stability by law. t-statistics are shown in parentheses; ***, ** and * denote statistical signicance
at the 1%, 5% and 10% levels, respectively. Fixed effects estimation.
FST index
FST components

(I)
General framework

(II)
FS coverage

MP transparency

0.05***
(7.00)
0.01***
(2.96)
0.07***
(2.70)
0.01
(0.20)
0.01
(0.52)
780
0.07

0.52***
(13.28)
0.06***
(7.62)
0.17
(1.13)
0.48
(1.23)
0.01
(0.20)
780
0.18

(III)
Tests and indicators

(V)
General framework

0.08***
(7.18)
0.02***
(9.26)
0.02
(0.60)
0.28**
(2.40)
0.01**
(2.11)
780
0.07

0.05***
(6.66)
0.01***
(3.22)
0.05*
(1.70)
0.04
(0.49)
0.0001
(0.93)
320
0.04

(VI)
FS coverage

Full sample

GDP p.c.
Banking crisis
Supervisory structure
Market capitalization
No. of observations
R-sqr. adj.

(VII)
Tests and indicators

Restricted sample
0.65***
(10.08)
0.04***
(3.58)
0.41*
(1.67)
0.01
(0.01)
0.01
(0.32)
320
0.19

0.09***
(5.30)
0.02***
(5.61)
0.06
(0.96)
0.53***
(3.64)
0.01**
(2.46)
320
0.07

Table 7
Correlation among FST index components. Note: All correlation coefcients are statistically signicantly different from 0 at 1% level.

General framework
Fin. stab. report coverage
Tests and Indicators
Website

General framework

Fin. stab. report coverage

Tests and indicators

Website

1
0.41
0.31
0.48

1
0.59
0.35

1
0.27

References
Andersson, M., 2008. Ten years with the nancial stability reports. Riksbank Econ.
Rev. 1.
Angeletos, G.-M., Pavan, A., 2007. Efcient use of information and social value of
information. Econometrica 75 (4), 11031142.
Blinder, A.S., Ehrmann, M., Fratzscher, M., de Haan, J., Jansen, D.-J., 2009. Central
bank communication and monetary policy: a survey of theory and evidence. J.
Econ. Lit. 46 (4), 910945.
Balakrishnan, R., Danninger, S., Elekdag, S., Tytell, I., 2009. The Transmission of
Financial Stress from Advanced to Emerging Economies., IMF Working Paper.
No.09/133.
Born, B., Ehrmann, M., Fratzscher, M., 2012. Macroprudential policy and central bank
communication. Int. Finance 15 (2), 179203.
Born, B., Ehrmann, M., Fratzscher, M., 2014. Central bank communication on nancial
stability. Econ. J. 577, 701734.

Breuer, T., Jandacka, M., Rheinberger, K., Summer, M., 2011. How to nd plausible,
severe, and useful stress scenarios. Int. J. Central Bank. 5 (3), 205224.
Cardarelli, R., Elekdag, S., Lall, S., 2011. Financial stress and economic contractions.
J. Financ. Stab. 7 (2), 7897.
Cihak, M., 2006. How Do Central Banks Write on Financial Stability?, IMF working
paper. No. 06-163.

Cihak, M., Munoz,

S., Teh Sharifuddin, S., Tintchev, K., 2012. Financial Stability
Reports: What Are They Good For?, IMF working paper. No. 12-1.
Claessens, S., Ghosh, S., Mihet, R., 2013. Macro-prudential policies to mitigate nancial system vulnerabilities. J. Int. Money Finance 39, 153185.
Cornand, C., Heinemann, F., 2008. Optimal degree of public information dissemination. Econ. J. 118, 718742.
Crowe, C., Meade, E.E., 2008. Central bank independence and transparency: evolution and effectiveness. Eur. J. Polit. Economy 24, 763777.
Crowe, C., 2010. Testing the transparency benets of ination targeting: evidence
from private sector forecasts. J. Monetary Econ. 57, 226232.

56

R. Horvth, D. Vasko / Journal of Financial Stability 22 (2016) 4556

Cukierman, A., 2009. The limits of transparency? Economic Notes 38 (1-2), 137.
Dale, S., Orphanides, A., sterholm, P., 2011. Imperfect central bank communication
- information versus distraction. Int. J. Central Bank. 7 (2), 339.
Dincer, N., Eichengreen, B., 2014. central bank transparency and independence:
updates and new measures. Int. J. Central Bank. 10 (1), 189253.
Ehrmann, M., Eijfnger, S., Fratzscher, M., 2012. The role of central bank transparency
for guiding private sector forecasts. Scand. J. Econ. 114 (3), 10181052.
Eijfnger, S., Geraats, P., 2006. How transparent are central banks? Eur. J. Polit.
Economy 22 (1), 121.
Frankel, J., Saravelos, G., 2012. Can leading indicators assess country vulnerability?
Evidence from the 200809 global nancial crisis. J. Int. Econ. 87 (2), 216231.
Franta, M., Barunik, J., Horvath, R., Smidkova, K., 2014. Are Bayesian fan charts useful
for central banks? Uncertainty, forecasting, and nancial stability stress tests.
Int. J. Central Bank. 10 (1), 159187.
Gerlach-Kristen, P., 2004. Is the MPCs voting record informative about future UK
Monetary Policy? Scand. J. Econ. 106, 299313.
Gick, W., Pausch, T., 2012. Persuasion by Stress Testing: Optimal Disclosure of Supervisory Information in the Banking Sectors. Deutsche Bundesbank, Discussion
Paper 32-2012.
Goldstein, I., Sapra, H., 2014. Stress tests and information disclosure. Found. Trends
Finance 8, 154.

Horvth, R., Smdkov, K., Zpal, J., 2012. Central banks voting record and future
policy. Int. J. Central Bank. 8 (4), 119.
Liedorp, F., Mosch, R., van der Cruisen, C., de Haan, J., 2013. Trans. Bank. Supervisors.
IMF Economic Review 61, 310335.
Meade, E.E., Stasavage, D., 2008. Publicity of debate and the incentive to dissent:
evidence from the US Federal Reserve. Econ. J. 118 (528), 695717.
Melecky, M., Podpiera, A.M., 2013. Institutional structures of nancial sector
supervision, their drivers and historical benchmarks. J. Financ. Stab. 9 (3),
428444.
Morris, S., Shin, H.S., 2002. Social value of public information. Am. Econ. Rev. 92 (5),
15211534.
Oosterloo, S., de Haan, J., Jong-a-Pin, R., 2007. Financial stability reviews: a rst
empirical analysis. J. Financ. Stab. 2 (4), 337355.
Siklos, P., 2011. Central bank transparency: another look. Appl. Econ. Lett. 18 (10),
929933.
van der Cruijsen, C., Eijfnger, S., Hoogduin, L., 2010. Optimal central bank transparency. J. Int. Money Finance 29 (8), 14821507.
Vazquez, F., Tabak, B.M., Souto, M., 2012. A macro stress test model of credit risk for
the Brazilian banking sector. J. Financ. Stab. 8, 6983.
Walsh, C.E., 2007. Optimal economic transparency. Int. J. Central Bank. 3 (1), 536.