The Architecture of Central Bank Transparency: Accounting Information and Financial Stability as Structural Pillars of Monetary Policy Transparency

Abstract

This article offers a structural reappraisal of central bank Monetary Policy Transparency (MPT) by explicitly incorporating two dimensions that have long remained peripheral in the literature: Accounting Information Transparency (AIT) and Financial Stability Transparency (FST). Building on a rigorous theoretical foundation, we develop two original transparency dimensions centered on AIT and FST, designed to extend a widely recognized monetary policy transparency index developed in the existing literature. This extension aims to capture, in an integrated manner, the institutional and macroprudential foundations that underpin the credibility, coherence, and effectiveness of modern monetary policy. The empirical analysis relies on a balanced panel of 25 countries over the period 2000–2019 and employs both Ordinary Least Squares (OLS) and the Generalized Method of Moments (GMM) to address potential endogeneity concerns and ensure the structural robustness of the estimations. The results provide strong evidence that both AIT and FST exert a positive, statistically significant, and economically meaningful effect on MPT. These findings substantially enrich the analytical framework of central bank transparency by demonstrating that high-quality financial reporting and transparent macroprudential communication constitute fundamental pillars of central banks’ credibility capital in an increasingly complex and globalized financial environment.

1. Introduction

Central bank transparency has, over the past two decades, established itself as a fundamental pillar of monetary credibility, expectations anchoring, and macro-financial stability. The seminal contributions of Eijffinger and Geraats (2006), followed by the updates provided by Dincer and Eichengreen (2009), Dincer et al. (2022), and, more recently, Acosta (2023) have offered a rigorous conceptualization of monetary policy transparency (MPT) as a multidimensional construct encompassing communication on policy objectives, analytical frameworks, decision-making procedures, and macroeconomic outlooks. However, despite the substantial expansion of this literature, the transparency architecture remains incomplete: two crucial components have received comparatively limited attention in both measurement and empirical assessment, namely the transparency of central banks’ accounting information and the transparency of financial stability communication.

The emergence of unconventional monetary policies since the Global Financial Crisis of 2008 has profoundly altered the structure and scale of central bank balance sheets, thereby amplifying the importance of accounting quality, financial risk disclosure, and prudential communication (Borio & Zabai, 2018; Bindseil, 2020; Zaini, 2025). At the same time, the rise in systemic risk concerns and the strengthening of macroprudential frameworks have conferred a central role upon financial stability transparency in shaping expectations and reducing informational asymmetries (Adrian et al., 2019). Nevertheless, existing MPT indices only marginally capture these two dimensions, leaving a substantial analytical gap in our understanding of contemporary monetary governance mechanisms. In other words, modern central bank transparency increasingly relies on balance-sheet disclosure and macroprudential communication, yet these dimensions remain insufficiently integrated into the standard empirical frameworks of monetary policy transparency.

From a theoretical perspective, this paper explicitly articulates three complementary channels through which AIT and FST can shape MPT. First, through an information channel, accounting and financial stability disclosures reduce informational asymmetries by improving the public’s understanding of balance-sheet exposures, risk profiles, and systemic vulnerabilities. Second, through a credibility and expectations channel, enhanced disclosure strengthens the coherence and predictability of central bank communication, thereby supporting expectations anchoring (Blinder et al., 2017; Kuang et al., 2025). Third, through a governance and accountability channel, accounting and prudential transparency increase auditability, reduce agency problems, and reinforce the legitimacy of monetary authorities, particularly when monetary policy and financial stability objectives interact.

This study seeks to fill this gap by developing two new transparency indices devoted respectively to Accounting Information Transparency (AIT) and Financial Stability Transparency (FST), and by empirically assessing their contribution to monetary policy transparency. Importantly, in the empirical strategy, AIT and FST are treated as explanatory dimensions influencing MPT; they are not defined as mechanical subcomponents of the benchmark MPT index used as the dependent variable, thereby avoiding circular reasoning.

From an empirical perspective, this study relies on a balanced panel of 25 countries over the period 2000–2019, thereby encompassing a wide range of institutional settings, balance-sheet structures, and monetary policy regimes. This observation window corresponds to the latest harmonized cross-country MPT data available in the updated reference dataset (Dincer et al., 2022), ensuring consistency and international comparability of the dependent variable. Methodologically, the analysis combines Ordinary Least Squares (OLS) estimations with dynamic panel System GMM estimations, which are well suited to settings characterized by persistence in transparency practices and allow for addressing potential endogeneity concerns, including reverse causality between transparency and institutional governance. This econometric design is intended to strengthen the robustness of inference in a setting where bidirectional relationships between transparency dimensions cannot be excluded.

The empirical results indicate that both AIT and FST exert a positive and statistically significant effect on MPT (Mamoon et al., 2025). Accounting information transparency strengthens institutional discipline and mitigates informational asymmetries related to the structure and risk profile of central bank balance sheets. In parallel, financial stability transparency enhances the credibility of monetary authorities by improving the public’s understanding of systemic risk prevention mechanisms. Taken together, these dimensions contribute to greater coherence in monetary announcements, improve policy predictability, and reinforce central banks’ capacity to anchor expectations.

Overall, this study contributes to the literature by proposing an original conceptual and empirical framework grounded in the construction of two novel transparency dimensions and by providing new evidence on their structuring role in monetary policy transparency. It also carries relevant policy implications by highlighting that improvements in accounting and prudential transparency constitute increasingly important levers for strengthening the credibility, legitimacy, and effectiveness of central banks.

The remainder of the article is organized as follows. Section 2 reviews the existing literature and develops the theoretical foundations of the study. Section 3 presents the methodology, the dataset, and the construction of the new transparency indices. Section 4 reports and discusses the empirical results. Finally, Section 5 concludes by highlighting the main implications for the design of central bank transparency.

2. Related Literature Review

The contemporary literature on central bank transparency has primarily focused on monetary policy communication in the narrow sense, particularly on the clarity of objectives, the predictability of interest-rate decisions, and the dissemination of macroeconomic information, while largely leaving aside two dimensions that are nevertheless central to modern monetary governance: Accounting Information Transparency (AIT) and Financial Stability Transparency (FST). This emphasis is consistent with the structure of standard Monetary Policy Transparency (MPT) measures, which mainly capture political, economic, procedural, policy, and operational transparency. In particular, Dincer et al. (2022) update the benchmark MPT index for 112 central banks through 2019, reinforcing the empirical centrality of “monetary-policy communication” in cross-country research. Yet, the quality of accounting information constitutes a fundamental pillar of the informational environment within which monetary authorities operate. By reducing information asymmetries and strengthening internal governance mechanisms, AIT enhances the efficiency of resource allocation and budgetary discipline (Biddle et al., 2009; Pinnuck & Lillis, 2007; García Lara et al., 2016), although some studies emphasize ambivalent effects associated with excessive or insufficient investment behavior (Bhattacharya et al., 2003). Beyond its intrinsic attributes such as accounting conservatism (Easley & O’hara, 2004) or the comparability of financial information (De Franco et al., 2011; Chen et al., 2012), AIT structures the entire informational framework within which monetary policy signals are generated, interpreted, and transmitted to economic agents. In this respect, AIT is not confined to a microeconomic determinant of institutional performance but directly conditions the credibility and effectiveness of MPT by facilitating the understanding of monetary decisions and reducing informational uncertainty (Yang et al., 2021). This argument is directly aligned with the information-asymmetry perspective: higher-quality disclosure reduces uncertainty, improves signal extraction, and strengthens market discipline mechanisms that are particularly relevant for policy institutions whose actions rely on expectations formation.

This relationship is all the more critical given that central banks are systemic institutions whose credibility also rests on the quality and completeness of the financial information they disclose. The literature thus emphasizes the importance of the regular and detailed publication of financial statements—namely the income statement, the balance sheet, and the cash flow statement—as essential instruments of transparency and accountability. The income statement makes it possible to assess institutional performance and the use of public resources, beyond the mere disclosure of profits that are largely conditioned by monetary policy decisions and the monopoly over currency issuance (Barth et al., 1998; Sullivan, 2005). The disclosure of operational costs associated with the conduct of monetary policy therefore emerges as a robust indicator of institutional effectiveness and as a factor limiting political interference, thereby indirectly contributing to central bank independence. Similarly, the balance sheet constitutes a central tool for analyzing the solvency, liquidity, and strategic orientations of monetary authorities, particularly in a context marked by the unprecedented expansion of central bank balance sheets following the implementation of unconventional monetary policies (Watts, 1974; Ou & Penman, 1989; Ohlson, 1995). Recent policy-oriented evidence highlights that balance-sheet policies can generate sizeable valuation and income effects, including episodes of losses, making transparent accounting explanations and disclosures a key condition for sustaining legitimacy and credibility. In this respect, central banks can preserve public trust by clearly explaining the drivers of reported losses and linking balance-sheet outcomes to mandates and policy choices. The cash flow statement complements this information by providing a dynamic reading of monetary transmission mechanisms, documenting the sources and allocation of financial resources, and revealing the trade-offs between monetary and budgetary objectives (Barth et al., 2001; Krishnan & Largay, 2000; Sullivan, 2005). Beyond academic work, international technical guidance dedicated to central banks explicitly stresses that IFRS consistent, transparent reporting strengthens cross-country comparability, governance, and accountability dimensions that conceptually underpin AIT as a transparency pillar.

In addition to financial statements, audit reports play a decisive role as a fundamental lever for enhancing the credibility of accounting information. Historically, auditing emerged in response to the need to reduce information asymmetries and conflicts of interest within complex institutional environments (DeAngelo, 1981). In the specific context of central banks, the publication of rigorous audit reports strengthens the reliability of financial statements, limits systemic risks associated with poor-quality information, and enhances market discipline (Craswell et al., 2002; Knechel et al., 2013, 2020). This certification function is particularly critical for MPT, insofar as it ensures consistency between stated objectives and the underlying financial reality, thereby optimizing the transmission of monetary signals to financial markets. From an agency-theory standpoint, independently audited and publicly disclosed accounts reduce agency costs by strengthening verifiability and accountability in institutions that combine high operational independence with delegated authority and significant balance-sheet discretion.

Moreover, the literature highlights the growing role of off-balance-sheet activities in transforming the bank credit channel and weakening the direct control of monetary authorities over liquidity conditions (Glick & Plaut, 1989; Altunbas et al., 2009; Loutskina, 2011). Off-balance-sheet commitments such as guarantees, credit commitments, or derivative instruments, although largely invisible, directly affect systemic risk and monetary transmission. Their opacity represents a major challenge for accounting information transparency and, by extension, for MPT, as it constrains the ability of both authorities and markets to accurately diagnose financial conditions. This point is increasingly salient in modern central banking where contingent liabilities, risk-sharing arrangements, and quasi-fiscal exposures may materially affect perceived policy sustainability; hence, AIT also improves “constraints transparency”, i.e., the observability of balance-sheet constraints that condition the credibility of policy commitments.

This transparency issue naturally extends to the literature on financial stability, which is now widely recognized as being closely intertwined with monetary policy (Nair & Anand, 2020). While the traditional consensus equated financial stability with price stability (Bordo et al., 2000), more recent studies have demonstrated that major financial imbalances may develop in low-inflation environments (Borio & Lowe, 2002; Borio, 2003). This debate has prompted deeper reflection on the role of central banks in preventing systemic risks and on the associated communication instruments. The publication of financial stability reports has thus emerged as a central transparency tool aimed at informing the public, strengthening institutional accountability, and improving coordination among authorities (Oosterloo et al., 2007; Svensson, 2003). These reports rely on the disclosure of financial soundness indicators, stress tests, and market-based indicators designed to cover the full spectrum of endogenous and exogenous risks affecting the financial system (Schinasi, 2006; Horváth & Vaško, 2016). However, the literature emphasizes significant heterogeneity in the content, clarity, and completeness of these reports, as well as the absence of fully harmonized standards (Čihák, 2006; Čihák et al., 2012). Finally, several authors stress the importance for central banks to explicitly define financial stability and its objectives as a prerequisite for any credible and measurable policy framework (Allen & Wood, 2006; Smaga, 2013; Rieu-Foucault, 2018). Recent empirical evidence confirms that financial stability reports can contain “news” for markets and affect market activity, supporting the view that FST is an informational policy instrument rather than a purely descriptive document (Harris & Roark, 2019). More broadly, recent research shows that financial stability communication (through reports or speeches) carries measurable informational content and can be linked to policy decisions and expectations formation (Istrefi et al., 2023). In addition, governance arrangements shape how financial stability communication is produced and how effective it is: Londono et al. (2021) develop a conceptual framework and show that governance frameworks influence communication strategies and their effectiveness in mitigating deteriorations in financial-cycle conditions.

Taken together, this literature reveals three persistent limitations that motivate the present study and clarify the research gap. First, most empirical work on central bank transparency continues to rely on standard MPT indices that are primarily designed around monetary-policy communication, leaving balance-sheet disclosure and macroprudential transparency only weakly represented (Dincer et al., 2022). Second, the accounting transparency literature, while extensive, typically does not embed central bank financial disclosure within the broader transparency architecture that shapes policy credibility, expectations, and communication effectiveness. Third, the financial stability communication literature documents heterogeneity in reporting practices and market effects but is rarely integrated into a unified transparency framework that links macroprudential disclosure to monetary policy transparency and credibility channels. As a result, the literature provides limited cross-country evidence on whether, how, and through which mechanisms accounting disclosure (AIT) and financial stability communication (FST) jointly contribute to monetary policy transparency in a systematic and measurable way.

In this context, the existing literature appears fragmented, as it examines AIT, FST, and MPT separately, without offering an integrated framework capable of capturing their dynamic interactions. It is precisely this gap that our contribution addresses by proposing original indices of AIT and FST, designed as structural determinants of MPT, both from a theoretical and an empirical perspective. Importantly, our conceptualization is designed to address the circularity concern raised in the transparency literature: AIT and FST are treated as complementary institutional transparency pillars that can shape MPT through identifiable channels, rather than being defined as mechanical components of the benchmark MPT index used as the dependent variable in the empirical analysis.

Grand theoretical framework. Building on the above evidence, the hypotheses are grounded in three complementary theoretical perspectives. (i) Information-asymmetry theory predicts that higher-quality accounting and macroprudential disclosure reduces uncertainty and strengthens the informational content of institutional signals, thereby improving the interpretability of monetary policy communication. (ii) Agency theory implies that verifiable and audited disclosure reduces agency costs and strengthens accountability in independent institutions with delegated authority and significant balance-sheet discretion. (iii) Credibility and expectations frameworks predict that clearer disclosure of objectives, constraints, risk assessments, and trade-offs improves expectations anchoring and policy predictability. These perspectives jointly provide a unified conceptual foundation linking AIT and FST to MPT and to the broader architecture of central bank credibility and legitimacy.

In light of this integrated review, two major theoretical mechanisms emerge. On the one hand, AIT shapes the informational environment within which monetary policy is designed and communicated. By improving the quality, comparability, and credibility of the financial information disclosed by central banks, particularly through the publication of financial statements, audit reports, and off-balance-sheet commitments, AIT reduces informational asymmetries, strengthens institutional discipline, and facilitates the interpretation of monetary signals by economic agents. In this framework, AIT does not merely constitute an instrument of financial accountability, but rather a structural determinant of MPT itself, insofar as it conditions the readability, coherence, and credibility of monetary communication. This mechanism further implies that transparency should not be reduced to policy statements alone but must also include disclosure about institutional constraints and balance-sheet exposures that condition the feasibility and credibility of policy commitments.

On the other hand, the literature highlights the growing entanglement between monetary policy and financial stability, especially in the context of rising systemic risks and the widespread adoption of unconventional monetary policies. The regular publication of financial stability reports, the explicit articulation of financial stability objectives, and the dissemination of financial soundness indicators contribute to reducing macro-financial uncertainty and enhancing the understanding of monetary policy trade-offs. In this sense, FST appears as a natural extension of the transparency architecture by reinforcing the coherence of the macro-financial governance framework and strengthening the transmission of monetary policy signals to financial markets. By clarifying systemic-risk assessments and policy trade-offs, FST strengthens expectations formation and supports credibility, particularly in periods of financial stress, consistent with evidence that such communications can affect market activity and policy reactions (Harris & Roark, 2019; Istrefi et al., 2023; Londono et al., 2021).

Based on these theoretical arguments and on the gaps identified in the existing literature, which largely addresses these dimensions in a fragmented and compartmentalized manner, this study formulates the following research hypotheses:

Hypothesis 1 (H1). 

Greater AIT of central banks exerts a positive and statistically significant effect on MPT.

Hypothesis 2 (H2). 

Greater FST exerts a positive and statistically significant effect on MPT.

These hypotheses reflect the central idea that MPT cannot be fully understood or adequately assessed independently of the quality of accounting information and the transparency of the financial stability framework. They constitute the analytical foundation for the construction of the new indices proposed in this article and guide the empirical analysis aimed at evaluating their structuring role in contemporary monetary governance.

3. Data and Methodology

In order to achieve the objectives of the study, this section has successively presented the sample, the construction of the two new transparency indices relating to AIT and FST, as well as the full set of explanatory and control variables employed in the empirical analysis. It has also set out the specification of the selected econometric models, combining ordinary least squares estimations with the generalized method of moments, so as to account simultaneously for unobserved heterogeneity and potential endogeneity issues. This integrated methodological framework thus provides a rigorous empirical foundation for assessing the structuring role of AIT and FST in shaping the MPT.

3.1. Sample

The empirical analysis relies on a balanced panel of 25 countries observed over the period 2000–2019. The sample includes both developed and emerging economies, thereby introducing substantial institutional, regulatory, and macro-financial heterogeneity. This heterogeneity is essential for identifying differentiated transparency structures and ensuring sufficient cross-sectional variation in disclosure practices.

The selection of countries is based on three methodological criteria:

(i)

The availability of sufficiently detailed and publicly accessible institutional documentation required for transparency coding;

(ii)

Cross-country comparability of macroeconomic and governance variables over the entire observation period;

(iii)

The existence of meaningful variation in monetary governance, accounting disclosure, and financial stability communication practices.

The complete list of countries included in the empirical analysis is presented in

Table 1. Sample composition and country coverage (2000–2019).

Group	Countries	Country List
Advanced economies	12	Australia, Canada, Denmark, Hong Kong SAR, Japan, New Zealand, Norway, Singapore, Sweden, Switzerland, United Kingdom, United States
Emerging economies	13	Brazil, Chile, China, Czech Republic, Egypt, Hungary, Indonesia, Kuwait, Mexico, Poland, Qatar, Russia, Saudi Arabia
Total	25	Balanced panel, annual frequency, 2000–2019

.

3.2. Data: Definition of Variables

3.2.1. Independent Variables

The construction of the new transparency dimensions proposed in this study fits strictly within the methodological continuity of the foundational approach developed by Eijffinger and Geraats (2006) and Dincer et al. (2022). We fully adopt their analytical logic, coding structure, unit of analysis, as well as the principles of information collection and processing, in order to ensure conceptual consistency and the comparability of results with existing transparency indices. Our approach therefore does not aim to substitute an alternative methodological framework, but rather to extend and enrich the initial index by integrating two dimensions that have hitherto been absent, yet have become essential in the contemporary architecture of monetary governance: AIT and FST.

Although AIT and FST are designed as structural dimensions of the broader transparency architecture, they are introduced separately in the empirical framework in order to identify their specific and marginal contribution to overall monetary policy transparency. This distinction preserves analytical clarity and avoids circular interpretation, as the benchmark measure of monetary policy transparency (MPT) remains that of Dincer et al. (2022), while AIT and FST are treated as additional components, of transparency.

The informational basis mobilized relies on an exhaustive corpus of public documents produced by central banks, in accordance with the sources used by Eijffinger and Geraats (2006). It notably includes annual financial reports, financial statements (balance sheets, income statements, and cash flow statements), audit reports, financial stability reports, as well as all institutional documents officially published on central bank websites. The use of these sources guarantees both the reliability, international comparability, and institutional legitimacy of the collected information, as these documents constitute the official channels of communication of monetary authorities and the preferred supports of institutional transparency.

From a methodological standpoint, we favor manual content analysis, in perfect alignment with the authors’ initial approach, rather than an automated analysis. This choice is justified by the heterogeneous and multilingual nature of the documents examined, a significant share of which is not fully available in the form of digitized data exploitable by standardized textual analysis software. The coding unit retained is the sentence, or even the word, exactly as that in the reference approach. Each occurrence is coded whenever it explicitly provides information relating to AIT or to FST, whether it concerns past, present, or future exposures, identified risks, financial outlooks, or management mechanisms implemented by the central bank.

The data are thus coded along two distinct but complementary axes: on the one hand, AIT, encompassing the publication of financial statements, audit reports, and off-balance-sheet commitments; on the other hand, FST, covering the publication of dedicated reports, the dissemination of indicators used in the analysis of systemic risks, as well as the explicit formalization of the definition and objectives of financial stability. By rigorously maintaining the methodology, information sources, and analytical structure of the founding authors, while extending the scope of transparency to these unprecedented institutional and prudential dimensions, our approach ensures at once methodological continuity, conceptual coherence, and analytical innovation.

The two new dimensions of transparency, namely accounting transparency and financial stability transparency, are based on a methodological approach derived from the monetary policy transparency index initially proposed by Eijffinger and Geraats (2006), as well as its subsequent extensions. The objective of this study is to enrich this reference framework by incorporating new dimensions that reflect current institutional and prudential issues, while preserving the conceptual and methodological foundations that have ensured the consistency and dissemination of the index in the literature. From this perspective, maintaining equal weighting between the dimensions appears to be a natural and consistent choice. The dimensions that make up the index, whether they come from the initial framework or are newly introduced, are designed to capture distinct and complementary facets of monetary policy transparency. In the absence of a theoretical or empirical consensus allowing for the establishment of an objective and stable hierarchy between these dimensions, the assignment of differentiated weightings would introduce a degree of subjectivity that would be difficult to justify and could compromise the internal balance of the derived index. By preserving the weighting logic adopted in the founding index, the MPTNWIX thus ensures methodological continuity, conceptual consistency, and comparability of results, while allowing for a gradual and harmonious expansion of the measurement of transparency. This approach ensures that the addition of new dimensions enriches the analysis without altering the fundamental structure of the index and promotes a robust and consistent interpretation of empirical results.

Accounting Information Transparency (AIT)

The transparency of central bank accounting information enables financial market participants to better understand how economic resources are measured and presented in financial statements. The detailed scoring methodology related to accounting disclosure is presented in Appendix A.

(a)

If the central bank publishes its financial statements (balance sheets, cash flow statements and profit and loss accounts), it is assigned a score between 0 and 1.5.

• If it publishes its balance sheets, it receives a score of ½; if not, it receives a score of 0.
• If it publishes its cash flow statements, it receives a score of 1/2; if it does not publish them, it receives a score of 0.
• If it publishes its profit and loss accounts, it is scored 1/2; otherwise it is scored 0.

(b)

If the central bank publishes audit reports they are given a score between 0 and 1/2.

• If they publish audit reports it receives a score of ½
• If not, it receives a score of 0.

(c)

If the central bank publishes its off-balance sheet activities, it is assigned a score between 0 and 1.

• If it publishes its off-balance sheet activities, it receives a rating of 1;
• If it does not publish its off-balance sheet activities, it is rated 0.

Financial Stability Transparency (FST)

Financial stability refers to the capacity of the financial system to remain resilient to economic shocks while continuing to perform its core intermediation and payment functions. The detailed operational definition and indicators used for scoring financial stability transparency are presented in Appendix A.

(a)

A score ranging from 0 to 1 is assigned depending on whether the central bank publishes Financial Stability Reports (FSRs).

• If it publishes FSRs, the score is 1;
• Otherwise, it is 0.

(b)

A score between 0 and 1 is assigned when Financial Stability Reports include a set of indicators used by the central bank in the analysis of financial stability.

• If so, a score of 1 is taken;
• Otherwise, a score of 0 is taken.

(c)

A score between 0 and 1 is assigned when Financial Stability Reports provide an explicit definition of financial stability together with clearly stated financial stability objectives.

• If yes, it is rated 1;
• If not, it is given a score of 0

3.2.2. Dependent Variable

• MPT Index

CBT refers to the absence of information asymmetry that exists between different economic agents and monetary policy makers. It is a multidimensional concept describing the extent to which central banks communicate information about the monetary policy-making process. According to Geraats (2002), the term MPT can be broken down into five aspects, namely political transparency, economic transparency, procedural transparency, policy transparency and operational transparency.

In analyzing the relevance of MPT, we rely on MPT data developed by Dincer et al. (2022). These data consist of an update of the original MPT index created by Eijffinger and Geraats (2006).

The empirical analysis relies on a balanced panel covering the period 2000–2019. The end date is determined by data availability and cross-country comparability considerations, as the benchmark Monetary Policy Transparency (MPT) measure employed in this study is taken from the updated Dinçer and Eichengreen dataset, which provides harmonized transparency scores consistently coded across countries up to 2019. Extending the observation window beyond 2019 would require a full re-coding of the MPT index from primary central-bank disclosures, including the reassessment of all five transparency dimensions using identical coding rules to preserve international comparability. Such an undertaking would constitute a distinct research project and falls beyond the scope of the present paper. Importantly, the period under study remains particularly appropriate for our research question, as it encompasses the Global Financial Crisis and the subsequent expansion of central bank balance sheets and macroprudential mandates developments that precisely motivated the growing relevance of accounting disclosure and financial stability communication within modern transparency architectures.

Appendix A provides an exhaustive documentation of the construction of the total index of monetary policy transparency, incorporating, beyond the five dimensions established in the literature, two novel dimensions specifically developed in this study. This structural extension of the index constitutes the fundamental methodological contribution of our research.

3.2.3. Control Variables

• GDP per capita:

Gross domestic product (GDP) per capita is an important indicator of economic performance and is often used as a general measure of average living standards or economic well-being, despite some acknowledged shortcomings. For example, average GDP per capita does not provide information on how GDP is distributed among different citizens. For example, average GDP per capita may be rising, but a very large number of people may be worse off if income inequality is also rising.

Moreover, in some countries there may be a large number of non-resident frontier or seasonal workers, or even inflows and outflows of property income, and both of these phenomena imply that the value of output is different from the income of residents, leading to an over- or underestimation of their standard of living, National Accounts at a Glance 2013, OECD Publishing.

• Openness:

Trade openness reflects the sum of exports and imports normalized by GDP. As Mishra (2007) and Lane and Milesi-Ferretti (2008) argue, bilateral equity investment is highly correlated with the underlying trade pattern. Indeed, through trade, investors are better able to access accounting and regulatory information on international markets and thus invest in international assets. Default risk is also reduced by closer trade integration. Trade transactions can also directly generate cross-border financial flows, including trade credits, export insurance and payment facilities.

• A banking crisis dummy:

A banking crisis is defined as a situation where a large part of a country’s banking sector is facing severe financial difficulties, marked in particular by a considerable loss of capital and government interventions necessary to prevent a systemic collapse, as reported by Laeven and Valencia (2013). Two major criteria are used by these authors to identify a systemic banking crisis:

Public intervention criterion: the crisis leads to significant intervention by public authorities to stabilize the situation of threatened banks. This may involve capital injections, guarantees on bank deposits and debts, nationalizations or restructuring plans.

Banking system failure criterion: A banking crisis is characterized by the failure of a large number of banks, or by a scale of failures that significantly affect the economy as a whole, with adverse repercussions on access to credit and financial stability.

• Financial crisis dummy:

1 = Banking crisis, 0 = No crisis

The definition and measurement of the variables used in the empirical analysis are presented in

Table 2. Definition of Variables.

Variable	Definition	Measurement	Source
MPT	Monetary Policy Transparency Index	Composite index (5 dimensions)	Dincer et al. (2022)
AIT	Accounting Information Transparency	Score based on publication of financial statements, audit reports, off-balance sheet disclosure	Authors’ coding
FST	Financial Stability Transparency	Score based on FSR publication, indicators, definition and objectives	Authors’ coding
GDP per capita	Economic development indicator	GDP per capita (constant prices)	World Bank
Openness	Trade openness	(Exports + Imports)/GDP	World Bank
Banking crisis	Systemic banking crisis dummy	1 = crisis, 0 = no crisis	Laeven and Valencia (2013)
Financial crisis	Crisis dummy	1 = crisis, 0 = no crisis	Laeven and Valencia (2013)

.

3.3. Model Presentation

In this section, we detail the econometric methodology adopted in our study, aiming to explore empirically the time and space varying nature of the validity of the two new core indices we have created, namely AIT and FST. We investigate how and to what extent these two new indices contribute to the performance and valuation of the central bank transparency index (monetary policy transparency index MPT) established by Eijffinger and Geraats (2006) and updated by Dincer et al. (2022) in terms of number of years.

The econometric framework is based on time-series and cross-sectional analysis, taking into account the nature and sources of the data described in the previous subsections. The empirical estimations were conducted using Stata version 15.

$${AITindex}_{i,t}=X_{i,t-j}\beta +{\alpha }_{i,t}+e_{i,t}$$

(1)

${AITindex}_{i,t}$ denotes our index of transparency of central bank accounting information for country i in time (year) t. The explanatory variables are $X_{i,t-j}$. For the basic model, we assume that j = 0. A robustness check with j = 3 is performed. As mentioned above, we include in our selection of explanatory variables a MPT index, GDP per capita, openness, a banking crisis and financial crisis dummy. To evaluate our Equation (1), we rely on the fixed effects estimator.

$${FSTindex}_{i,t}=X_{i,t-j}\beta +{\alpha }_{i,t}+e_{i,t}$$

(2)

${FSTindex}_{i,t}$ denotes our central bank financial stability transparency index for country i in time (year) t. The explanatory variables are: $X_{i,t-j}$. For our baseline model, we assume j = 0. A robustness check with j = 3 is performed. As outlined earlier, our list of explanatory variables includes a MPT index, GDP per capita, openness, a banking crisis and financial crisis dummy. For the estimation of our Equation (2), we use the fixed effects estimation method.

4. Empirical Analysis

This section describes the model specification and estimation results.

4.1. Descriptive Statistics and Correlation Analysis

As one of the first steps in our estimation, we are going to carry out descriptive statistics, which encompass a vast array of methods for describing numerous sets of data using appropriate tools, and extracting the essential information that results.

4.1.1. Descriptive Statistics

Table 3. Descriptive statistics of the variables AIT.

Variable	Obs	Min	Max	Mean	Std. Dev	Skewness	Kurtosis
MPT	500	1	14.5	8.55	3.466632	−0.2879023	2.147918
AIT	500	0	3	2.016	0.867322	−1.211464	3.380013
lnGDPPCAP	500	6.617747	11.54164	9.895552	1.110101	−0.8744506	3.030371
lnOPSS	500	2.973466	6.092711	4.285676	0.657109	0.7596728	3.806638
BKGCRS	500	0	1	0.056	0.230152	3.862184	15.91646
FCIALCRS	500	0	1	0.65	0.477447	−0.6289709	1.395604

Note: MPT is an index of monetary policy transparency; AIT is the index of transparency of accounting information; lnGDPPCAP is the Logarithm of Gross Domestic Product per capita; lnOPSS is the logarithm of the openness; BKGCRS is the banking crisis variable; FCIALCRS is a financial crisis dummy variable.

and

Table 4. Descriptive statistics of the variables FST.

Variable	Obs	Min	Max	Mean	Std. Dev	Skewness	Kurtosis
MPT	500	1	14.5	8.55	3.466632	−0.2879023	2.147918
FST	500	0	3	1.918	1.228941	−0.6801721	1.820652
lnGDPPCAP	500	6.617747	11.54164	9.895552	1.110101	−0.8744506	3.030371
lnOPSS	500	2.973466	6.092711	4.285676	0.657109	0.7596728	3.806638
BKGCRS	500	0	1	0.056	0.230152	3.862184	15.91646
FCIALCRS	500	0	1	0.65	0.477447	−0.6289709	1.395604

Note: MPT is an index of monetary policy transparency; FST is an index of financial stability transparency; lnGDPPCAP is the Logarithm of Gross Domestic Product per capita; lnOPSS is the logarithm of the openness; BKGCRS is the banking crisis variable; FCIALCRS is a financial crisis dummy variable.

provide the statistics of the different variables for our sample, during the period from 2000 to 2019 for 500 observations, examining the minimum (Min), maximum (Max), mean, standard deviation (SD), median (Median), skewness and kurtosis. Statistics are provided for the entire sample, for developed countries, as well as for emerging countries.

Table 3 presents the descriptive statistics of the variables used in the empirical analysis, based on a balanced panel of 500 observations (25 countries over 2000–2019).

The Monetary Policy Transparency index (MPT) exhibits a mean value of 8.55, with a minimum of 1 and a maximum of 14.5, indicating substantial cross-country variation in transparency levels. The relatively high standard deviation (3.47) confirms pronounced dispersion across institutional environments. The distribution is slightly negatively skewed (−0.29), suggesting a moderate concentration of observations at higher levels of transparency. The kurtosis (2.15), below the benchmark value of 3 for a normal distribution, indicates a mildly platykurtic distribution, implying a somewhat flatter distribution with limited extreme concentration.

The Accounting Information Transparency index (AIT) displays a mean of 2.02 on a scale ranging from 0 to 3, reflecting a relatively elevated but heterogeneous level of accounting disclosure practices across central banks. The negative skewness (−1.21) indicates that a significant share of observations lies toward the upper bound of the index, consistent with widespread publication of core financial statements and audit reports among the sampled institutions. The kurtosis (3.38) suggests a moderate concentration around the mean with some clustering at higher transparency levels.

Regarding the control variables, the logarithm of GDP per capita (lnGDPPCAP) presents a mean of 9.90 and a standard deviation of 1.11, reflecting considerable heterogeneity in economic development levels across the sample. The negative skewness (−0.87) indicates a relative concentration of middle- and high-income economies.

Trade openness (lnOPSS) records a mean of 4.29 and a standard deviation of 0.66. The positive skewness (0.76) suggests that a subset of highly open economies contributes to the right tail of the distribution, while maintaining overall dispersion across countries.

The crisis dummy variables reveal distinct distributional characteristics. The banking crisis variable (BKGCRS) exhibits a low mean (0.056), confirming the relative rarity of systemic banking crises during the sample period. Its high positive skewness (3.86) and elevated kurtosis (15.92) indicate that crisis episodes are infrequent but concentrated in a limited number of observations. In contrast, the broader financial crisis variable (FCIALCRS) records a mean of 0.65, suggesting that crisis periods according to this broader definition occur in a substantial proportion of the sample. The lower kurtosis (1.40) indicates a flatter distribution relative to normality.

Overall, the descriptive statistics confirm substantial cross-country and temporal variability across transparency, macroeconomic, and crisis indicators. This variability provides sufficient dispersion to justify panel estimation techniques and supports the relevance of dynamic specifications addressing potential endogeneity concerns.

Since both specifications (the model including AIT and the model including FST) rely on the same dataset (25 countries over the period 2000–2019, totaling 500 observations) and employ the same dependent variable (MPT) as well as an identical set of control variables, the descriptive statistics related to MPT and the controls remain strictly unchanged from one table to another. Consequently, in order to avoid any redundancy, we focus exclusively on the new variable of interest, namely FST.

Table 4 reports that FST exhibits a mean value of 1.918 on a scale ranging from 0 to 3, suggesting an intermediate level of financial stability transparency across the sampled central banks. The standard deviation (1.229) indicates substantial cross-country dispersion, confirming meaningful heterogeneity in financial stability disclosure practices.

The distribution displays moderate negative skewness (−0.68), indicating that observations are relatively concentrated toward higher levels of FST, with fewer instances at the lower end of the scale. The kurtosis value (1.82), below the benchmark value of 3 for a normal distribution, suggests a mildly platykurtic distribution, reflecting dispersion without excessive concentration around the mean.

Overall, the variability observed in FST confirms the presence of sufficient cross-sectional differentiation to support its inclusion in the subsequent panel estimations. Given that both empirical specifications rely on the same balanced dataset and identical control variables, differences in the results across models can be attributed to the specific transparency dimension introduced.

4.1.2. Correlation Analysis

To check the multicollinearity between the independent and control variables, we first use the Pearson two-by-two correlation matrix (

Table 5. Pearson correlation matrix (AIT).

Full Sample
	MPT	AIT	lnGDPPCAP	lnOPSS	BKGCRS	FCIALCRS
MPT	1.000
AIT	0.5148	1.000
lnGDPPCAP	0.3737	0.4029	1.000
lnOPSS	−0.0025	0.3369	0.2453	1.000
BKGCRS	0.0932	0.0106	0.0431	−0.0447	1.000
FCIALCRS	0.2643	0.0692	0.2772	0.0466	0.0875	1.000

Note: MPT is an index of monetary policy transparency; AIT is the index of transparency of accounting information; lnGDPPCAP is the Logarithm of Gross Domestic Product per capita; lnOPSS is the logarithm of the openness; BKGCRS is the banking crisis variable; FCIALCRS is a financial crisis dummy variable.

and

Table 6. Pearson correlation matrix (FST).

	MPT	FST	lnGDPPCAP	lnOPSS	BKGCRS	FCIALCRS
MPT	1.000
FST	0.5457	1.000
lnGDPPCAP	0.3737	0.2673	1.000
lnOPSS	−0.0025	0.1450	0.2453	1.000
BKGCRS	0.0932	−0.0475	0.0431	−0.0447	1.000
FCIALCRS	0.2643	0.3916	0.2772	0.0466	0.0875	1.000

Note: MPT is an index of monetary policy transparency; FST is an index of financial stability transparency; lnGDPPCAP is the Logarithm of Gross Domestic Product per capita; lnOPSS is the logarithm of the openness; BKGCRS is the banking crisis variable; FCIALCRS is a financial crisis dummy variable.

). This matrix gives us the correlation coefficient between the independent variables. Gujarati (2002) indicates that these coefficients must be less than 0.8 among the independent variables, otherwise there will be a serious problem of multicollinearity. The objective of this bivariate analysis is to verify the absence of multicollinearity between the explanatory variables, namely the independent variable (AIT) and (FST), the dependent variable (MPT) and the control variables (lnGDPPCAP, lnOPSS, BKGCRS, FCIALCRS), in order to guarantee the robustness of the estimates in the regression models.

Table 5 reports the Pearson correlation matrix. In accordance with Gujarati’s (2002) recommendations, a correlation between independent variables greater than 0.80 could indicate a problem of collinearity. Examination of the correlation matrix reveals that the majority of correlations between variables are below this threshold, suggesting an absence of multicollinearity. In conclusion, the correlation matrix analysis shows that the variables used in the model do not exhibit sufficiently high levels of correlation to raise concerns about multicollinearity, which ensures the robustness of the estimates in the regression models considered, in line with the recommendations of Gujarati (2002).

Table 6 presents the Pearson correlation matrix. As shown in Table 6, none of the pairwise correlation coefficients exceeds the conventional threshold of 0.80, indicating that multicollinearity does not pose a serious concern in the empirical specification.

4.2. Results and Regression Analysis

In this section, we examine whether the impact of AIT on MPT as well as that of FST on MPT has changed at the global level. We therefore use interaction terms to study this effect. An interaction effect occurs when a relationship between (at least) two variables is modified by (at least) one other variable.

To better understand the economic dynamics at play, we propose a two-dimensional econometric analysis. This analysis is divided into a static analysis to establish immediate relationships, followed by a dynamic analysis to capture temporal adjustments and long-term effects.

4.2.1. Ordinary Least Squares (OLS): Classical Econometric Analysis and Preliminary Results

In this first phase, the analysis focuses on exploring the direct relationships between AIT and MPT, as well as between FST and MPT, while incorporating relevant control variables such as GDP per capita, economic openness, and financial and banking crises. The aim is to shed light on the underlying mechanisms linking these economic variables, without taking account of temporal dynamics, which is useful for identifying direct effects, thus providing an initial interpretation of economic relationships in a given context.

Table 7. Static panel estimation: MCO results (AIT).

Variables	Coefficient
AIT	2.029001 (0.000) ***
lnGDPPCAP	0.5280945 (0.000) ***
lnOPSS	−1.167503 (0.000) ***
BKGCRS	0.8162654 (0.126)
FCIALCRS	1.363641 (0.000) ***
Const.	3.305209 (0.010) **
Prob > Chi2 = 0.0000

Note: MPT is an index of monetary policy transparency; AIT is the index of transparency of accounting information; lnGDPPCAP is the Logarithm of Gross Domestic Product per capita; lnOPSS is the logarithm of the openness; BKGCRS is the banking crisis variable; FCIALCRS is a financial crisis dummy variable. *, **, ***: significant at 10%, 5% and 1%, respectively. The p-values are reported between parentheses.

reports the results of the static panel estimation (OLS) examining the relationship between Accounting Information Transparency (AIT) and Monetary Policy Transparency (MPT). The coefficient associated with AIT is positive and statistically significant at the 1% level (β = 2.029; p < 0.01), indicating a strong and economically meaningful positive association between accounting disclosure practices and the overall level of MPT.

From a theoretical perspective, this result is consistent with the literature emphasizing that high-quality financial information enhances institutional credibility and reduces informational asymmetries (Bushman & Smith, 2003; Barth et al., 1998). More recent contributions on central bank communication further stress that transparency improves expectations management and strengthens policy effectiveness, particularly in environments characterized by heightened uncertainty (Blinder et al., 2017; Coibion et al., 2019). By extending this reasoning to the domain of accounting disclosure, the present finding suggests that financial reporting transparency contributes to structuring the informational environment within which monetary policy decisions are interpreted by financial markets and economic agents. In this sense, AIT complements the multidimensional transparency framework initially formalized by Geraats (2002) and subsequently updated by Dincer et al. (2022).

Economically, the magnitude of the coefficient implies that improvements in accounting disclosure including the publication of balance sheets, income statements, cash flow statements, audit reports, and off-balance-sheet commitments are associated with substantial increases in measured MPT. This is particularly relevant in the post-Global Financial Crisis context, during which central bank balance sheets expanded considerably and financial positions became increasingly intertwined with monetary policy implementation (Borio, 2019; Adrian et al., 2019). Transparent financial reporting may therefore enhance the coherence between announced policy objectives and the underlying financial constraints of the institution.

Among the control variables, the logarithm of GDP per capita is positive and statistically significant (β = 0.528; p < 0.01), in line with evidence that higher levels of economic development are systematically associated with stronger institutional transparency (Dincer et al., 2022). Trade openness (lnOPSS) enters with a negative and statistically significant coefficient (β = −1.168; p < 0.01). While openness reflects integration into global markets (Lane & Milesi-Ferretti, 2008), this result suggests that greater external exposure does not automatically translate into higher formal transparency in monetary communication, possibly reflecting heterogeneous institutional responses to external constraints.

The financial crisis dummy (FCIALCRS) is positive and statistically significant (β = 1.364; p < 0.01), indicating that crisis episodes are associated with higher levels of MPT. This finding is consistent with the view that central banks intensify communication during systemic stress in order to stabilize expectations and restore credibility (Crowe & Meade, 2008; Blinder et al., 2017). By contrast, the banking crisis dummy is not statistically significant, suggesting that sector-specific distress alone does not systematically induce structural changes in overall MPT within the static specification.

The joint significance of the regressors (Prob > Chi² = 0.000) confirms the explanatory relevance of the model. Nevertheless, given the static nature of the estimation, the coefficients should be interpreted as conditional associations rather than definitive causal effects. Potential endogeneity concerns including reverse causality between transparency dimensions and MPT are explicitly addressed in the subsequent dynamic panel estimations using the Generalized Method of Moments.

Table 8. Static panel estimation: MCO results (FST).

Variables	Coefficient
FST	1.41465 (0.000) ***
lnGDPPCAP	0.8483646 (0.000) ***
lnOPSS	−0.7222596 (0.000) ***
BKGCRS	1.506586 (0.005) **
FCIALCRS	−0.0712592 (0.802)
Const.	0.4989867 (0.686)
Prob > Chi2 = 0.0000

Note: MPT is an index of monetary policy transparency; FST is an index of financial stability transparency; lnGDPPCAP is the Logarithm of Gross Domestic Product per capita; lnOPSS is the logarithm of the openness; BKGCRS is the banking crisis variable; FCIALCRS is a financial crisis dummy variable. The significance levels (two-tail test) are * = 10%, ** = 5% and *** = 1%.

presents the static panel (OLS) estimation results for the specification including Financial Stability Transparency (FST). The coefficient associated with FST is positive and statistically significant at the 1% level (β = 1.415; p < 0.01), indicating a strong positive conditional association between financial stability transparency and the Monetary Policy Transparency (MPT) index.

This finding is consistent with the growing literature emphasizing the increasing interdependence between monetary policy and financial stability functions in modern central banking (Borio, 2019; Adrian et al., 2019). As macroprudential mandates have expanded following the Global Financial Crisis, communication regarding systemic risks, stress-testing frameworks, and financial stability objectives has become progressively intertwined with monetary policy communication. In line with Dincer et al. (2022), transparency is multidimensional and evolving; the present result suggests that enhanced disclosure in the financial stability domain is systematically associated with higher levels of monetary transparency.

From an economic standpoint, the magnitude of the coefficient implies that improvements in financial stability disclosure such as the publication of Financial Stability Reports, explicit risk indicators, and formalized definitions of stability objectives are associated with substantial increases in measured MPT. However, it is important to emphasize that, within the static OLS framework, this relationship captures conditional correlation rather than definitive causality. Potential reverse causality and unobserved institutional heterogeneity are explicitly addressed in the subsequent dynamic GMM estimations.

Among the control variables, the logarithm of GDP per capita remains positive and highly significant (β = 0.848; p < 0.01), reinforcing the well-established link between economic development and institutional transparency (Dincer et al., 2022). Trade openness (lnOPSS) enters with a negative and statistically significant coefficient (β = −0.722; p < 0.01). While openness reflects international integration (Lane & Milesi-Ferretti, 2008), its negative association with MPT suggests that external exposure does not automatically translate into greater formal transparency in monetary communication, a result that warrants further investigation in the dynamic framework.

Regarding crisis variables, the banking crisis dummy (BKGCRS) is positive and statistically significant (β = 1.507; p < 0.01), indicating that episodes of banking instability are associated with higher levels of MPT within the static specification. This result is consistent with evidence that central banks tend to intensify communication during periods of financial stress to stabilize expectations and restore credibility (Blinder et al., 2017; Weber, 2019). In contrast, the financial crisis dummy (FCIALCRS) is not statistically significant, suggesting that broader crisis classifications do not uniformly translate into measurable adjustments in MPT once other controls are included.

The overall joint significance of the regressors (Prob > Chi² = 0.000) confirms the explanatory relevance of the model. Taken together, these results provide preliminary evidence of a strong institutional association between financial stability transparency and monetary policy transparency, supporting the view that macro-financial disclosure practices are closely aligned within contemporary central banking frameworks. The robustness of this relationship is further evaluated using dynamic panel techniques to account for potential endogeneity.

The estimation by OLS constitutes a relevant and necessary methodological step in empirical analysis, insofar as it allows for a clear and intuitive identification of contemporaneous relationships between the variables of interest. Within the framework of this study, the initial use of OLS provides a first assessment of the effect of AIT and FST on MPT, while controlling for a coherent set of macroeconomic and institutional factors. This approach thus offers a robust static reading of the underlying structural correlations and enables verification of the directional consistency and economic significance of the relationships under investigation, in line with standard empirical practices in the literature on monetary governance.

However, MPT is inherently a dynamic phenomenon, characterized by strong temporal persistence and potentially endogenous interactions with its institutional and macro-financial determinants. In this context, a strictly static estimation may prove insufficient to fully capture intertemporal adjustment mechanisms and may lead to biased estimates, particularly in the presence of endogeneity arising from reverse causality, omitted variables, or the inclusion of a lagged dependent variable. In other words, while OLS constitutes an indispensable informative step, it alone does not allow for a rigorous isolation of the dynamic and causal effects of AIT and FST on MPT.

In particular, given the well-documented persistence of institutional transparency indicators and the potential bidirectional relationship between transparency dimensions and monetary policy credibility, the adoption of a dynamic GMM framework is essential to mitigate dynamic panel bias and to strengthen the identification strategy beyond simple contemporaneous associations.

It is precisely in order to overcome these methodological limitations without calling into question the validity of the initial static results that this study subsequently adopts a dynamic panel approach based on the Generalized Method of Moments (GMM). The use of GMM makes it possible to explicitly incorporate the dynamic dimension of MPT by introducing lags of the dependent variable and the explanatory variables, while endogenously addressing potential issues of simultaneity and unobserved heterogeneity. By relying on appropriate internal instruments, the GMM framework provides a more robust identification of causal relationships and enables the analysis of long-term interactions between AIT, FST, and MPT within a realistic institutional and macroeconomic setting.

Thus, this methodological choice fully justifies the transition from the OLS model to the GMM model, not as a rejection of the static estimations, but as their natural and necessary extension. It ensures the econometric robustness of the results, strengthens their empirical credibility, and allows for a more comprehensive and convincing analysis of the structural determinants of MPT.

4.2.2. Advanced Estimation Using Generalized Method of Moments (GMM): Robustness and Precision of Results

To ensure the econometric robustness and causal validity of our dynamic panel estimations, we rely on the GMM, which is particularly suited to settings characterized by persistence, endogeneity, and unobserved heterogeneity. This framework exploits orthogonality conditions between lagged variables and the error term and is implemented through two alternative estimators: the difference GMM estimator proposed by Arellano and Bond (1991) and the system GMM estimator developed by Blundell and Bond (1998).

The system GMM estimator is particularly appropriate in this context, as it improves efficiency in the presence of highly persistent dependent variables and mitigates potential weaknesses of lagged levels as instruments in the difference equation.

The validity of the instrument set is systematically assessed using the Sargan (1980) and Hansen (1982) tests of over-identifying restrictions, whose null hypothesis stipulates that the instruments are exogenous. To prevent instrument proliferation and the associated weakening of over-identification tests, the instrument matrix is carefully restricted and collapsed where appropriate, ensuring that the number of instruments remains below the number of cross-sectional units.

The empirical results indicate that the corresponding p-values consistently exceed conventional significance thresholds, thereby supporting the validity of the lagged variables used as instruments. In addition, the reliability of the dynamic specification is evaluated through the Arellano Bond tests for serial correlation in the residuals. While first-order autocorrelation is expected by construction, the absence of second-order autocorrelation (AR (2)) confirms the appropriateness of the instrument strategy and the overall consistency of the GMM estimates. Taken together, these diagnostic tests provide strong empirical support for the robustness and internal coherence of the dynamic panel estimations reported in this study.

While the dynamic GMM framework substantially strengthens identification by mitigating simultaneity bias and accounting for persistence, the estimated coefficients should be interpreted as strengthened causal inferences under maintained assumptions rather than definitive structural proof.

The results in

Table 9. Dynamic panel estimation: GMM system results (AIT).

Variables	Coef.	p > |t|
L1.MPT	0.8595484	(0.000) ***
AIT	0.1988271	(0.000) ***
lnGDPPCAP	−0.3521351	(0.000) ***
lnOPSS	0.0441113	(0.875)
FCIALCRS	0.2378398	(0.000) ***
BKGCRS	−0.1980451	(0.014)
cons	4.088156	(0.000) ***
Arellano–Bond test for AR (1) in first differences: z = −3.82 Pr > z = 0.000
Arellano–Bond test for AR(2) in first differences: z = −0.99 Pr > z = 0.323
Hansen test of overid. restrictions: Chi2(31) = 22.46 Prob > Chi2 = 0.868
Sargan test of overid. restrictions: Chi2(31) = 443.01 Prob > Chi2 = 0.000

Note: MPT is an index of monetary policy transparency; AIT is the index of transparency of accounting information; lnGDPPCAP is the Logarithm of Gross Domestic Product per capita; lnOPSS is the logarithm of the openness; BKGCRS is the banking crisis variable; FCIALCRS is a financial crisis dummy variable. *, **, ***: significant at 10%, 5% and 1%, respectively. The p-values are reported between parentheses.

and

Table 10. Dynamic panel estimation: GMM system results (FST).

MPT	Coef	p > |t|
MPT L1.	0.81297	(0.000) ***
FST	0.0837902	(0.001) ***
lnGDPPCAP	−0.5485465	(0.000) ***
lnOPSS	0.3761881	(0.252)
BKGCRS	−0.1559997	(0.160)
FCIALCRS	0.3259905	(0.001) ***
_cons	5.210831	(0.008) **
Arellano–Bond test for AR(1) in first differences: z = −3.80 Pr > z = 0.000
Arellano–Bond test for AR(2) in first differences: z = −0.91 Pr > z = 0.362
Hansen test of overid. restrictions: Chi2(29) = 21.94 Prob > Chi2 = 0.823
Sargan test of overid. restrictions: Chi2(29) = 438.25 Prob > Chi2 = 0.000

Note: MPT is an index of monetary policy transparency; FST is the index of transparency of financial stability; lnGDPPCAP is the Logarithm of Gross Domestic Product per capita; lnOPSS is the logarithm of the openness; BKGCRS is the banking crisis variable; FCIALCRS is a financial crisis dummy variable. *, **, ***: significant at 10%, 5% and 1%, respectively. The p-values are reported between parentheses.

show that Arellano and Bond’s (1991) AR (1) and AR (2) levels of auto-correlation of order 1 and 2 essentially verify the hypothesis of autocorrelation of the residuals. Indeed, the residuals obtained are supposed to be correlated to order 1, since their probabilities are lower than 10%, but not to order 2, since their probabilities are higher than 10%. This is true for our data.

The results displayed in Table 7 and Table 8 show that the results remain broadly similar to our main estimates.

The results of the GMM-system model are shown in the following Table 9 and Table 10.

The diagnostic statistics reported in Table 9 provide strong support for the validity and internal consistency of the dynamic specification. The Arellano–Bond test for first-order serial correlation in first differences is significant (z = −3.82, p = 0.000), as expected in dynamic panel models. Crucially, the AR(2) test does not reject the null hypothesis of no second-order serial correlation (z = −0.99, p = 0.323), thereby satisfying the key consistency condition of the system GMM estimator (Arellano & Bond, 1991; Blundell & Bond, 1998).

The Hansen test of over-identifying restrictions further confirms the validity of the instrument set (χ²(31) = 22.46, p = 0.868), indicating that the null hypothesis of instrument exogeneity cannot be rejected (Hansen, 1982; Roodman, 2009). The relatively high p-value does not suggest overfitting but rather supports the adequacy of the collapsed instrument matrix used in the estimation. In contrast, the Sargan test is rejected (p = 0.000), which is not unexpected under heteroskedasticity and does not invalidate the system GMM estimator, as the Hansen statistic provides the robust criterion for instrument validity in two-step estimations. Taken together, these diagnostics reinforce the credibility of the identification strategy and mitigate concerns regarding endogeneity and reverse causality raised in previous critiques.

The coefficient of the lagged dependent variable (L1.MPT = 0.8595, p < 0.01) reveals a very high degree of persistence in monetary policy transparency. This magnitude implies that nearly 86% of past transparency levels carry over into the current period, highlighting strong institutional inertia and path dependency. Economically, this suggests that reforms in transparency frameworks generate gradual and cumulative effects rather than immediate structural shifts, thereby fully justifying the adoption of a dynamic specification (Acemoglu & Robinson, 2006; Blundell & Bond, 1998).

After controlling for this persistence and addressing potential simultaneity bias, Accounting Information Transparency (AIT) retains a positive and highly significant coefficient (0.1988, p < 0.01). This result confirms that improvements in accounting disclosure practices are systematically associated with higher levels of monetary policy transparency (Zaini, 2025). In economic terms, a one-unit increase in AIT leads to an approximate 0.20 increase in MPT, ceteris paribus, which represents a non-negligible structural effect given the bounded nature of transparency indices. Importantly, the stability of this coefficient relative to static estimations suggests that the relationship is not merely contemporaneous but structurally embedded within institutional governance mechanisms (La Porta et al., 1998; Djankov et al., 2008). While the GMM framework strengthens causal inference by mitigating endogeneity, the interpretation remains conditional on the maintained validity of internal instruments rather than constituting definitive structural causality (Roodman, 2009).

The coefficient of GDP per capita (−0.3521, p < 0.01) is negative and statistically significant in the dynamic specification. This reversal compared to static models suggests that once institutional persistence is accounted for, higher levels of economic development may be associated with more complex policy environments in which transparency practices evolve under competing stabilization and communication objectives. This finding highlights the importance of dynamic adjustment mechanisms and cautions against purely contemporaneous interpretations (Eichler & Littke, 2018; Weber, 2019; Aftab & Mehmood, 2023).

Economic openness (lnOPSS) is statistically insignificant (p = 0.875), indicating that its effect is largely absorbed by institutional persistence and structural determinants. This suggests that openness operates as a contextual characteristic rather than as a long-run structural driver of transparency once dynamic feedback effects are controlled (Rajan & Zingales, 2003; Mishkin, 2004).

Crisis variables exhibit asymmetric effects. Financial crises (FCIALCRS) have a positive and highly significant impact (0.2378, p < 0.01), consistent with post-crisis reform dynamics and credibility-restoration strategies through enhanced communication (Eichengreen, 2000; Geraats, 2002; Weber, 2019). By contrast, banking crises (BKGCRS) exert a negative and significant effect (−0.1980, p = 0.014), suggesting temporary opacity or discretionary communication under acute systemic stress (Diamond & Dybvig, 1986; Aftab & Mehmood, 2023). These contrasting effects underscore the complex interaction between crisis management and transparency regimes.

Overall, the system GMM results confirm that AIT constitutes a structurally significant and dynamically robust determinant of monetary policy transparency. By explicitly addressing persistence, endogeneity, and instrument validity, the dynamic specification substantially strengthens the empirical credibility of the findings while avoiding overstatement regarding causal interpretation.

The dynamic panel results reported in Table 10 satisfy the standard identification and consistency conditions of the two-step system GMM estimator (Arellano & Bond, 1991; Blundell & Bond, 1998). The Arellano–Bond test detects first-order serial correlation in first differences (p = 0.000), which is expected in dynamic specifications, while the absence of second-order serial correlation (AR(2), p = 0.362) confirms the validity of the moment conditions. The Hansen test of over-identifying restrictions does not reject the null hypothesis of instrument exogeneity (p = 0.823), supporting the adequacy of the internal instrument set (Hansen, 1982; Roodman, 2009). Although the Sargan statistic is rejected, this test is not robust under heteroskedasticity and therefore does not invalidate inference based on the Hansen criterion. Taken together, these diagnostics indicate that the dynamic specification appropriately addresses persistence, unobserved heterogeneity, and potential simultaneity bias.

The coefficient of the lagged dependent variable (L1.MPT = 0.8129, p < 0.01) reveals a high degree of persistence in monetary policy transparency. Economically, this magnitude implies that transparency behaves as a strongly path-dependent institutional attribute, evolving through gradual adjustments rather than abrupt structural breaks. This finding is consistent with the dynamic nature of transparency frameworks documented in recent central banking research (Geraats, 2014; Binder, 2017) and fully justifies the adoption of a dynamic estimation strategy.

Controlling for persistence and potential reverse causality, Financial Stability Transparency (FST) exhibits a positive and statistically significant coefficient (0.0838, p = 0.001). Although moderate in magnitude, the effect is economically meaningful given the bounded scale of transparency indices. The result indicates that improvements in the disclosure of systemic risks, macroprudential instruments, and financial stability objectives are associated with higher levels of monetary policy transparency (Silva et al., 2025). This finding supports the growing body of literature emphasizing the increasing institutional interdependence between monetary and macroprudential communication frameworks in the post-global financial crisis era (Masciandaro et al., 2011; Horváth & Vaško, 2016; Lombardi & Siklos, 2016; Londono et al., 2021). More recent empirical contributions further document that enhanced financial stability communication contributes to improved policy predictability and reduced macro-financial uncertainty (Eichler & Littke, 2018; Weber, 2019; Aftab & Mehmood, 2023; Fadda et al., 2025). The persistence of the FST coefficient within the dynamic framework strengthens the credibility of this relationship by mitigating simultaneity concerns; however, consistent with best practice in dynamic panel estimation, the interpretation remains conditional on the maintained validity of the instrument set rather than implying definitive structural causality.

Regarding control variables, the logarithm of GDP per capita displays a negative and highly significant coefficient (−0.5485, p < 0.01). Once institutional persistence and endogeneity are controlled for, higher levels of economic development appear associated with relatively smaller incremental gains in transparency. This may reflect increasing institutional sophistication in advanced economies, where credibility is increasingly anchored in established reputational capital, policy track records, and institutional maturity rather than solely in disclosure intensity (Woodford, 2005; Geraats, 2014; Bholat et al., 2019).

Economic openness (lnOPSS) does not exert a statistically significant effect (p = 0.252), suggesting that the transparency openness nexus is not dynamically robust across heterogeneous institutional settings. This result aligns with evidence indicating that the impact of international integration on transparency depends on exchange rate regimes, capital flow management strategies, and institutional configurations (Eichler & Littke, 2018; Aftab & Mehmood, 2023).

With respect to crisis variables, banking crises do not generate a statistically significant dynamic effect once persistence is accounted for. In contrast, global financial crises exert a positive and significant impact (0.3260, p = 0.001), suggesting that systemic shocks are associated with subsequent transparency-enhancing institutional adjustments. This pattern is consistent with post-crisis reform dynamics documented in the literature on central bank communication and institutional credibility (Crowe & Meade, 2008; Weber, 2019; Dincer et al., 2022).

Overall, the system GMM findings indicate that Financial Stability Transparency constitutes a positive and dynamically robust determinant of Monetary Policy Transparency. By explicitly accounting for persistence, unobserved heterogeneity, and potential reverse causality, the dynamic specification enhances the empirical credibility of the estimated relationship while maintaining appropriate caution in causal interpretation.

5. Conclusions

This study contributes to a renewed and more comprehensive understanding of central bank transparency by extending the traditional analytical framework beyond the five classical dimensions of monetary policy transparency. While existing indices have predominantly focused on political, economic, procedural, policy, and operational transparency, our analysis demonstrates that accounting information transparency (AIT) and financial stability transparency (FST) constitute structural components of contemporary transparency architectures. Their integration provides a more institutionally coherent and operationally relevant measure of transparency in modern central banking.

From an empirical perspective, the results obtained from a balanced panel of 25 countries over the period 2000–2019 reveal that both AIT and FST exert positive and statistically significant effects on monetary policy transparency. These findings indicate that transparency is not confined to communication strategies regarding policy decisions but is deeply rooted in the broader informational environment within which central banks operate. The disclosure of comprehensive financial statements, audit reports, and off-balance-sheet commitments enhances institutional accountability and reduces informational asymmetries. Similarly, structured communication regarding systemic risks and clearly articulated financial stability objectives strengthens the overall credibility and internal coherence of transparency practices.

Beyond statistical significance, the results carry substantive economic meaning. A one-unit increase in AIT or FST corresponds to the disclosure of additional core transparency components, reflecting a tangible reinforcement of institutional information availability. This suggests that transparency improvements are not merely symbolic but are associated with concrete enhancements in disclosure practices that shape market expectations and policy credibility.

The study also highlights the multidimensional and cumulative nature of transparency. Monetary policy transparency cannot be fully understood independently of the accounting and financial stability frameworks that condition the sustainability and perceived credibility of policy actions. By incorporating AIT and FST into the transparency architecture, the proposed framework addresses an important limitation of existing indices and provides a more complete representation of institutional disclosure practices.

From a theoretical standpoint, the findings contribute to transparency and institutional credibility theories by demonstrating that informational asymmetry reduction operates not only through policy communication but also through accounting disclosure and macroprudential communication channels. Transparency thus emerges as a systemic institutional characteristic rather than a narrow communication tool, reinforcing the interdependence between monetary governance and financial stability structures.

From a policy perspective, the results suggest that strengthening monetary policy transparency requires coordinated improvements in accounting disclosure standards and financial stability communication practices. Enhancing the clarity, accessibility, and reliability of financial information may improve policy predictability, facilitate expectation anchoring, and reinforce public trust in monetary authorities. These implications are particularly relevant in institutional environments characterized by expanded central bank balance sheets and increased macroprudential responsibilities.

Finally, while the empirical framework is applied to a heterogeneous sample combining advanced and emerging economies, future research may further explore whether the magnitude of the transparency effects differs according to the level of institutional development or financial system maturity. Such extensions would contribute to a more refined understanding of how transparency architectures evolve across diverse macroeconomic contexts.