Next Article in Journal
From Fields to Finance: Dynamic Connectedness and Optimal Portfolio Strategies Among Agricultural Commodities, Oil, and Stock Markets
Previous Article in Journal
The Impact of Green Bond Issuance on Corporate Environmental and Financial Performance: An Empirical Study of Japanese Listed Firms
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

An X-Ray Using NLP Techniques of Financial Reporting Quality in Central and Eastern European Countries

by
Tatiana Dănescu
1 and
Roxana Maria Stejerean
2,*
1
Department of Economic Sciences, Faculty of Economics and Law, University of Medicine, Pharmacy, Science and Technology “George Emil Palade” of Targu Mures, 540566 Târgu Mures, Romania
2
Doctoral School of Accounting, “1 Decembrie 1918” University of Alba Iulia, 510009 Alba Iulia, Romania
*
Author to whom correspondence should be addressed.
Int. J. Financial Stud. 2025, 13(3), 142; https://doi.org/10.3390/ijfs13030142
Submission received: 11 June 2025 / Revised: 23 July 2025 / Accepted: 4 August 2025 / Published: 6 August 2025

Abstract

This study assesses the quality of financial reporting in ten Central and Eastern European countries using a methodology based on natural language processing (NLP) techniques. 570 annual reports of companies listed on the main index on the stock exchanges of 10 Central and Eastern European (CEE) countries, over the period 2019–2023, were evaluated to determine the degree of convergence of the following four measurable qualitative characteristics: relevance, exact representation, comparability and understandability. The main objective is to identify consistency in the quality of accounting information based on the application of an international financial reporting framework. The applied methodology eliminates subjective variability by implementing a standardized scoring system, aligned with the criteria developed by NiCE, using libraries such as spaCy and NLTK for term extraction, respective sentiment analysis and word frequency evaluation. The results reveal significant heterogeneity in all characteristics examined, with statistical tests confirming substantial differences between countries. The investigation of relevance revealed partial convergence, with three dimensions achieving complete uniformity, while the exact representation showed the highest variability. The assessment of comparability showed a significant difference between countries’ extreme values, and in terms of comprehensibility a formalistic approach was evident, with technical dimensions outweighing user-oriented aspects. The overall quality index varied significantly across countries, with a notable average deterioration in 2023, indicating structural vulnerabilities in financial reporting systems. These findings support initial hypotheses on the lack of homogeneity in the quality of financial reporting in the selected region, despite the implementation of international standards.

1. Introduction

In a global economic environment marked by increasing complexity and interdependence, the quality of financial information is becoming essential for the efficient functioning of capital markets. The literature emphasizes that financial reporting goes beyond the mere presentation of accounting data, representing a complex communication system that facilitates economic decision-making.
The contemporary financial reporting environment is characterized by the duality of its dimensions—external and internal—generating complex information dynamics that require sophisticated processing and analysis systems. In this context, the quality of financial information becomes a factor contributing to the reduction in information asymmetry, as well as to the optimization of the decision-making process.
To develop this scientific study, based mainly on previous research, along with current literature, the proposed research methodology is an extension of the investigative approach entitled “Companies’ behavior in measuring the quality of financial reports: Pre- and post-pandemic research” (Dănescu & Stejerean, 2022).
The preliminary study focused on a small sample, both geographically and in terms of size, together with the application of a manual predominant methodology to assess the qualitative criteria of financial reporting. We aim through this research to overcome the limitations identified above, by adopting a multidimensional methodological approach.
The process of automating the assessment of qualitative criteria is based on content analysis, i.e., on the use of natural language processing technologies, with the main purpose of objectivizing and streamlining the process of assessing the quality of financial reporting, facilitating the elimination of the subjectivity inherent in manual implementations and ensuring the reproducibility of the investigative results.
In today’s environment, marked by the continued expansion of global financial markets and the increasing complexity of economic transactions, the quality of financial information is particularly important in facilitating the process of ensuring efficiency and transparency of capital markets.
Qualitative characteristics, as defined in the General Conceptual Framework for Financial Reporting, constitute one of the theoretical underpinnings useful in understanding and assessing the usefulness of identified financial information in decision-making (Whittington, 2019). Relevance and faithful representation are considered fundamental characteristics, while comparability, verifiability, timeliness, and understandability are amplifying characteristics that amplify the usefulness of financial information to users (International Accounting Standards Board [IASB], 2018).
The theoretical foundations for the assessment of quality characteristics have been established by Van Beest, Braam and Boelens, who developed the Nijmegen Centre for Economics (NiCE) methodology, a comprehensive measurement tool for the comprehensive assessment of the quality of financial reporting. The NiCE index, consisting of 21 items measuring the qualitative characteristics of financial information, has been validated through inter-rater reliability tests, together with internal consistency, and has proven to be a robust tool for measuring the quality of financial reporting (Beest et al., 2009).
The current body of literature highlights a constant preoccupation with assessing the resulting effects on the quality of financial reporting because of the adoption of international standards. In this context, seminal research by Yurisandi and Puspitasari demonstrated that the adoption of IFRS has led to a significant increase in the quality of financial reporting, while also signaling a negative effect on the timeliness of information, mainly caused by the increased complexity in IFRS disclosure requirements (Yurisandi & Puspitasari, 2015).
Further studies in this direction, such as those conducted by Wahyuni et al., have reinforced the findings of the before mentioned study by conducting a meta-analysis of the literature to summarize the effects of IFRS implementation in Indonesia, while confirming the general trend of improvement in the quality of financial reporting because of the adoption of international standards (Wahyuni et al., 2020). Subsequently, other researchers., emphasized the positive impact of IFRS adoption on certain characteristics of financial reporting, such as faithful representation and comparability.
The methodology developed by NiCE has been adapted and applied in diverse economic contexts, allowing comparative assessments across different geographical regions and stages of economic development. For example, Kabwe (2023) assessed the quality of financial reporting in a developing country, highlighting significant shortcomings in the relevance and comparability of information, as well as a preference for historical cost valuation over fair value.
In the banking sector, Nguyen adapted the NiCE methodology to assess the financial reporting quality of commercial banks in Vietnam, demonstrating the versatility of the instrument in specific sector contexts (Nguyen, 2023) Similarly, Eliza used the same methodology to analyze the impact of digitization on the quality of financial reporting, finding significant improvements as a result of the implementation of digital accounting systems (Eliza, 2023).
Although the NiCE methodology has been widely used in Western Europe, North America and South East Asia, there are a limited number of studies investigating the quality of financial reporting in the specific context of Eastern European economies. The particularities of this region, particularly those related to the transition from planned to market economies and the relatively recent adoption of international standards, provide opportunities for further research.
The traditional methodology of applying the NiCE index involves manual analysis of financial reports raising issues of efficiency and scalability. In the existing literature, there is a lack of automated tools for calculating the NiCE index, which could facilitate its application on larger sample sizes and reduce the inherent subjectivity.
This research aims to fill the above-mentioned gaps by extending the application of the NiCE methodology in the specific context of Eastern European economies, thus contributing to the geographical diversification of research in this field, as well as to the development and validation of an algorithm for automating the calculation of the index, allowing the efficient processing of a large volume of financial reports.

2. Results

H1. 
There Is No Significant Convergence of Results on the Relevance of Reported Financial Information Between the Sample Countries.
Table 1 presents the descriptive statistical parameters for the four dimensions of relevance (R1–R4), aggregating data from 2019–2023.
Dimension R1 stands out as the only component that shows variability within this qualitative characteristic, registering a mean value of 1.16 and a coefficient of variation of 10.34%. This dimension demonstrates moderate differences between the analyzed countries, with values ranging between 1.00 and 1.50, suggesting different levels of implementation of the specific aspects of relevance captured by this component.
In contrast, dimensions R2, R3 and R4 exhibit complete uniformity across all countries and time periods analyzed, with a constant value of 1.00 and zero coefficients of variation (0.00%). This homogeneity indicates full convergence for these components of relevance across all jurisdictions studied.
Analysis of the aggregate scores by country (Table 2) reveals a limited stratification of the sample: (1) top performers: Croatia and Romania (1.08 and 1.05, respectively); (2) average performance: the Czech Republic, Lithuania, Estonia and Latvia (between 1.02 and 1.03); and (3) basic performance: Poland, Hungary and Bulgaria (1.00–1.02). The difference between the highest (Croatia: 1.08) and lowest (Bulgaria, Hungary: 1.00) value is about 8% of the overall average value, indicating a small gap between countries.
Intra-entity variability, as measured by the coefficient of variation, is generally low, except for Bulgaria (CV = 5.83%), which shows a one-off increase in 2022. Most countries show high stability over time (Lithuania and Poland: CV = 0.00%).
Dimension R1 shows the strongest correlation with the average relevance score, being the only component that contributes to the differentiation of the countries. Dimensions R2, R3 and R4, having constant values, do not contribute to the variability of the scores.
The results of the F-test for differences between the mean values of the countries indicate a statistically significant difference, but of low magnitude, providing empirical support for a partial convergence of the level of relevance between the countries analyzed. Convergence is almost complete for dimensions R2, R3 and R4, while dimension R1 shows moderate differences between countries, contributing to the limited heterogeneity observed at the aggregate level.
Based on the results of the analysis, the hypothesis on the absence of significant convergence is partially rejected, as almost complete convergence is observed for dimensions R2, R3 and R4 (zero variability), while only dimension R1 shows moderate differences between countries (CV = 10.34%). Therefore, the data supports the existence of a predominant convergence with limited elements of divergence, with the maximum difference between entities representing only 8% of the overall mean value.
H2. 
There Is No Significant Convergence of Results Across Sample Countries in Terms of Accurate Representation of Reported Financial Information.
Table 3 and Table 4 presents the descriptive statistical parameters for the five dimensions of credibility (F1–F5), aggregating data from 2019–2023.
Dimension F2 stands out as the dominant component of this qualitative characteristic in all the jurisdictions studied, with the highest mean value (3.76), significantly higher than the other dimensions. However, the high variability of this dimension (CV = 18.82%) indicates notable differences in the degree of implementation of standards across the countries analyzed, suggesting different levels of maturity in the adoption of international regulatory frameworks.
In contrast, dimensions F3, F4 and F5 show mean values close to the theoretical minimum (1.06, 1.01 and 1.02, respectively) and small coefficients of variation (11.83%, 5.84% and 4.58%, respectively), suggesting high uniformity across countries for these components of credibility. Dimension F1 shows the most pronounced variability (CV = 28.64%), with values ranging from 1.00 to 2.41, reflecting fundamental differences between countries in this aspect of credibility.
Analysis of the aggregate scores by country (Table 5 and Table 6) reveals a stratification of the sample into three distinct groups: (1) high performers: Lithuania, Estonia, Poland, Romania and Croatia (>1.75); (2) average performers: Bulgaria (1.69); and (3) moderate performers: Hungary, Latvia and the Czech Republic (<1.52). The difference between the highest (Lithuania: 1.80) and the lowest (Czech Republic: 1.44) is about 20% of the overall average value, indicating a moderate but statistically significant gap.
This stratification is also confirmed by the result of the F-test for differences in the mean values of the countries, which indicates a statistically significant difference (F = 7.36, p < 0.001), thus providing empirical support for the rejection of the hypothesis of complete similarity in the level of accurate representation between the countries analyzed.
Intra-entity variability, as measured by the coefficient of variation, also shows notable differences: countries with high stability (Czech Republic: CV = 1.35%, Romania: CV = 3.30%) versus countries with more pronounced fluctuations (Latvia: CV = 11.24%).
The F2 dimension shows the strongest correlation with the mean accurate representation score (r = 0.91), suggesting that this component has the most significant contribution to the differentiation of countries. The substantial variation in scores for F2 (CV = 18.82%) and the major discrepancies between countries (from 2.76 for the Czech Republic to 4.45 for Poland) highlight fundamental differences in the approach to compliance with reporting standards.
Dimension F1 presents a moderate correlation with the mean score (r = 0.58), but the highest variability (CV = 28.64%), suggesting that this component also contributes significantly to the differentiation of countries. The dimensions F3, F4 and F5 show weak or even negative correlations with the mean score (r = −0.22, r = 0.06, r = 0.06 and r = 0.21, respectively) and low variability, indicating a limited contribution to the differentiation of countries.
The cross-dimensional correlation matrix (Table 7) provides additional information on the interrelations between the components of credibility.
The generally weak correlations between dimensions (most below 0.3 in absolute value) suggest a relative independence of these components, indicating that they capture distinct aspects of the credibility of financial reporting.
The moderate positive correlation between F3 and F4 (r = 0.319) suggests a link between these components, possibly derived from their common grounding in the principles of consistency and transparency of financial reporting. The positive correlation between F2 and F5 (r = 0.284) reflects the complementarity between adherence to standards and the ability to independently verify financial information.
Based on the results of the analysis, the hypothesis on the absence of meaningful convergence is fully confirmed, with the F-test showing statistically significant differences between countries (F = 7.36, p < 0.001) and a clear stratification into three performance groups with a gap of about 20% between the extreme values. Heterogeneity is emphasized by the high variability of the F1 (CV = 28.64%) and F2 (CV = 18.82%) dimensions, which contribute significantly to the differentiation of entities in terms of accurate representation of financial information.
H3. 
The Structure and Content of the Financial Reports Published in the Selected Countries Do Not Ensure an Adequate Level of Comparability.
Statistical examination of the data on the comparability of financial statements provides an empirical basis for the assessment of Hypothesis 3, which postulates a similar degree of comparability between the analyzed countries. Table 8 presents the descriptive statistical parameters for the comparability criteria (C1–C6), aggregating data from 2019–2023.
Criteria C2 and C3 show the highest mean values (3.84 and 3.91) and low coefficients of variation (9.48% and 4.80%), indicating a high and relatively uniform level of compliance with accounting standards and timeliness in the presentation of financial information.
In contrast, criteria C1, C4 and C5 have substantially lower mean values (2.06, 2.14 and 2.66) and high coefficients of variation (27.18%, 24.77% and 25.11%), suggesting significant discrepancies between countries in these dimensions of comparability. Criterion C6 shows the highest variability (CV = 28.12%), with values ranging from 1.15 to 5.00, highlighting fundamental differences in the approach to consistency in the application of accounting policies between the countries analyzed.
There is a clear stratification of the countries in the sample into three distinct groups: (1) high performers, as shown in Table 9 and Table 10: Romania and Lithuania (>3.40); (2) medium performers: Croatia, Poland, Estonia and Hungary (2.90–3.30); and (3) moderate performers: Czech Republic, Bulgaria and Latvia (<2.80). The difference between the highest (Romania: 3.44) and the lowest (Latvia: 2.45) represents about 33% of the overall average value. The F-test for differences between the mean values of the entities indicates a statistically significant difference (F = 5.83, p < 0.001), providing strong empirical support for the acceptance of Hypothesis 3 regarding the lack of similarity in the degree of comparability between the analyzed countries and entities.
Intra-entity variability, as measured by the coefficient of variation, also shows differences: countries with high stability (Romania: CV = 2.64%, Czech Republic: CV = 3.42%) versus countries with pronounced fluctuations (Bulgaria: CV = 16.31%).
Dimensions C4 and C5 show the strongest correlations with the average comparability score (r = 0.92 and r = 0.93, respectively), suggesting that these components make the most significant contribution to the differentiation of countries. The wide variation in the scores for these criteria (CV = 24.77% and 25.11%) highlights fundamental differences in the approach to cross-sector and cross-national comparability.
Dimensions C2 and C3 show weak correlations with the mean score (r = 0.29 and r = 0.28, respectively), but the overall high levels for these dimensions (mean C2 = 3.84, mean C3 = 3.91) suggest relatively uniform implementation of these aspects of comparability across the jurisdictions analyzed.
Dimension C1 presents a strong correlation with the mean score (r = 0.75), but low overall values (mean = 2.06), indicating a significant contribution to the differentiation of countries but also a poor overall implementation of this aspect of comparability.
The correlation matrix between dimensions (Table 11) provides additional information on the interrelations between the components of comparability.
The strong positive correlation between C2 and C3 (r = 0.801) emphasizes the in-built link between standards adoption and temporal consistency, suggesting that countries that rigorously implement standards also tend to maintain high consistency over time.
The moderate to strong positive correlations between C4, C5 and C6 (r between 0.518 and 0.685) indicate an integrated approach to these dimensions, suggesting that countries that exceed well in one of these aspects tend to perform well in the others.
H4. 
The Financial Reports Published in the Selected Companies and Countries Do Not Ensure an Adequate Level of Intelligibility in the Presentation of Financial Information.
Examination of the empirical results on the level of understandability of financial information in the sample analyzed reveals complex and nuanced configurations. Table 12 presents the descriptive statistics for the five dimensions of comprehensibility assessed (U1–U5), aggregating data from 2019–2023.
Dimension U2 registers the highest average score (4.238), with values ranging from 2.600 to 5.000, indicating a consistent concern for transparency and clarity of financial disclosure in all jurisdictions analyzed.
U3 also shows a high mean score (3.543), but accompanied by the widest variability (CV = 24.73%), reflecting substantial differences across jurisdictions in the granularity of financial information provided. In contrast, the U1, U4 and U5 di-measures U1, U4 and U5 have substantially lower mean values (1.247, 1.201 and 1.198, respectively), indicating significant deficiencies in the user-focus of financial disclosures. This discrepancy between the technical (clarity, detail) and user-oriented dimensions suggests a predominantly formalistic approach to financial reporting, focusing on technical compliance at the expense of actual accessibility.
The stratification of countries (Table 13 and Table 14) is evident, with a group of high performers (Poland, Romania, Croatia, Lithuania and Hungary, with average values between 2.415 and 2.531) and a group of moderate performers (Estonia, Czech Republic, Bulgaria and Latvia, with average values between 1.885 and 2.129). The difference between the extreme values (Poland vs. Latvia) represents about 34% of the overall average value.
The F-test for differences between the mean values of the countries indicates a statistically significant difference (F = 4.273, p = 0.002), providing strong empirical support for the rejection of Hypothesis 4 regarding the similarity of the level of comprehensibility between the analyzed countries. Intra-entity variability, as measured by the coefficient of variation, also shows notable differences: countries with high stability (Croatia: CV = 1.55%, Lithuania: CV = 2.07%) versus countries with pronounced fluctuations (Bulgaria: CV = 12.70%).
The U3 dimension shows the strongest correlation, as shown in Table 15, with the mean intelligibility score (r = 0.927), suggesting that this component has the most significant contribution to the differentiation of countries. The wide variation in U3 scores (CV = 24.73%) and the major discrepancies between countries (from 1.903 for Latvia to 4.573 for Romania) highlight fundamental differences in the approach to the granularity of the information provided.
The dimensions U1 and U5 show moderate correlations with the mean score (r = 0.251, respective r = 0.312), but low overall levels for these dimensions (mean U1 = 1.247, mean U5 = 1.198) suggest insufficient concern for user orientation across the jurisdictions analyzed.
Dimension U4 shows a negative correlation with the mean score (r = −0.257), indicating a possible inverse relationship between technical complexity and accessibility of information.
The strong negative correlation between U2 and U4 (r = −0.730) emphasizes the in-built tension between technical clarity and practical accessibility of information. The moderate positive correlation between U1 and U3 (r = 0.547) indicates that countries that provide more detailed information also tend to provide better overall accessibility.
Based on the results of the analysis, the hypothesis is fully confirmed, with the F-test showing statistically significant differences between countries (F = 4.273, p = 0.002) and a clear stratification with a gap of about 34% between extreme values. Major weaknesses in user-orientation are highlighted by the low scores of dimensions U1, U4 and U5 (below 1.25), indicating a predominantly formalistic approach to financial reporting, focusing on technical compliance at the expense of effective user accessibility.
H5. 
At Global Level, the Financial Reports Do Not Demonstrate a Uniform Level of Quality.
Examination of the descriptive statistical parameters in the table below reveals aspects of the distribution and variability in the quality of financial reporting in the regional context analyzed (Table 16). There is a dispersion of the index values across the analyzed countries, with the coefficient variation for each year ranging from 7.69% (2021) to 11.49% (2020), with an average of 9.02% for the whole period, indicating a lack of homogeneity in the quality of financial reporting.
The difference between the maximum and minimum values of the index ranges between 1.58 points (2021) and 2.68 points (2023), representing between 19.5% and 33.0% of the average value of the index; this significant amplitude contravenes the uniformity assumption.
In terms of stratification, a grouping of countries into three distinct categories according to the average level of the index is identified: countries with a higher level (>8.50) represented by Romania (8.76), Lithuania (8.66), Croatia (8.59) and Poland (8.51); countries with an intermediate level (7.50–8.50) including Hungary (7.98) and Estonia (7.91); and countries with a lower level (<7.50) such as Bulgaria (7.33), the Czech Republic (7.32) and Latvia (6.81). The analysis of the coefficient of variation by year (Table 17) indicates a trend towards convergence in 2021 (CV = 7.69%), followed by divergence in subsequent years, suggesting that the standardization of reporting quality is a non-linear process, susceptible to contextual influences.
The analysis of the annual rates of change in the quality index allows the identification of significant trends in the quality of financial reporting. There is a re-reduction in the quality index for seven of the nine countries analyzed in 2023 compared to 2022, with an average rate of deterioration of 7.07%, this synchronization of the negative evolution suggesting the presence of exogenous factors with a transnational impact on the quality of financial reporting.
The magnitude of the deterioration in 2023, ranging between 5.95% and 16.65% for the affected countries, substantially exceeds the magnitude of the improvements recorded in previous years, indicating a structural vulnerability of the quality of financial reporting to disruptive factors. There are also substantial differences in the long-term evolution of the index across the analyzed countries, with the compound annual growth rate (CAGR) for the period 2019–2023 ranging between −2.50% for Poland and 1.46% for Bulgaria, which highlights the absence of a uniform pattern of evolution.
The pronounced volatility of annual rates of change for some countries, such as Bulgaria and Hungary, contrasts with the relative stability of others, such as Romania and Estonia, suggesting significant differences in the robustness of financial reporting systems and their resilience to disruptive factors.
For a more nuanced assessment of the uniformity hypothesis, we calculated the matrix of Pearson correlation coefficients between the index values for the analyzed countries, presented in Table 18.
The results of the correlational analysis reveal distinct patterns of interdependence in the evolution of the quality of financial reporting between the countries studied. We identify groups of countries with strongly correlated developments, structured in three main clusters: the first cluster comprises Croatia, Hungary, Romania and Estonia, with correlation coefficients ranging between 0.68 and 0.94, indicating a substantial synchronization of the reporting quality developments; the second cluster is represented by the Czech Republic, which shows strong negative correlations with the countries in the first cluster, suggesting a diametrically opposite dynamic; and the third cluster is Lithuania, which shows weak or insignificant correlations with the other countries.
The presence of significant negative correlations, such as that between the Czech Republic and Croatia (−0.93), points to the existence of specific endogenous factors that generate opposite dynamics in the quality of financial reporting, contradicting the uniformity hypothesis. Lithuania shows correlation coefficients close to zero with most of the other countries, suggesting an independent evolution of the quality of financial reporting, probably determined by specific contextual factors.
The average value of the correlation coefficients in absolute value is 0.55, indicating a moderate level of synchronization of the temporal evolution of the index, incompatible with the assumption of uniformity of the quality of financial reporting at the regional level.
Hypothesis 5, which postulates the existence of significant variations in the quality of financial reporting at the global level, despite the implementation of an international reference framework, is confirmed by the empirical results obtained which demonstrate that there is no convergence and uniformization of financial reporting standards.

3. Discussion

With the present study we contribute to the current academic literature in particular by applying large-scale automated natural language processing techniques to assess the quality of financial reporting by applying the NiCE methodology in the CEE region. In contrast to previous studies that relied on manual analysis of small samples (typically 20–50 companies), our automated framework allowed the analysis of 570 annual reports, increasing both the comprehensiveness and objectivity of the assessment, overcoming the scalability limitations identified in previous studies using the same methodology, and eliminating the subjective variability associated with manual assessments.
The empirical results obtained from a detailed investigation of 570 annual reports from entities listed on the main stock exchange indices in ten Central and Eastern European countries provide a nuanced and complex perspective on the quality of financial reporting in this developing region, revealing both notable progress in certain dimensions of quality and persistent and substantial challenges in the effective and uniform implementation of international financial reporting standards.
The detailed analysis of the relevance of financial information from the perspective of the four dimensions assessed reveals a complex and seemingly contradictory pattern of partial convergence, which contrasts significantly with the predominantly optimistic results reported in previous studies for other geographical regions and economic contexts. In contrast to the findings of Yurisandi and Puspitasari (2015), who document substantial and homogeneous improvements in the relevance of financial reporting following the adoption of International Financial Reporting Standards in the Indonesian economic context, the Central and Eastern European region presents a considerably more fragmented and heterogeneous picture, characterized by near-perfect convergence in some aspects of relevance and pronounced divergence in others.
Our findings highlight a different pattern from previous studies that have mainly focused on IFRS adoption. While Yurisandi and Puspitasari (2015) documented substantial improvements in Indonesia and Wahyuni et al. (2020) confirmed similar trends in Southeast Asia, our results for the CEE region demonstrate persistent heterogeneity despite decades of IFRS implementation. This divergence suggests that the benefits of adopting international standards are also significantly moderated by institutional and cultural factors specific to post-communist transition economies.
The dimension relating to the presence and quality of forward-looking statements and future-oriented information that contribute to the formation of expectations and forecasts regarding the future evolution of the entity shows the highest variability among the countries analyzed, with a coefficient variation of 10.34%, highlighting substantial differences in the approach to providing predictive information to facilitate the decision-making process of users of financial statements. Croatia and Romania emerge as the regional leaders in this critical dimension of relevance, with scores of 1.33 and 1.19, respectively, probably reflecting a higher degree of maturity of the national capital markets and a more intense pressure from institutional investors and financial analysts for information with higher predictive value.
The fundamental characteristic of accurate representation of financial information shows the most pronounced and worrying heterogeneity of all the quality characteristics analyzed in the present study, with F-test results indicating statistical differences between countries (F = 7.36, p < 0.001) revealing the existence of fundamental and structural differences in accounting culture.
The dimension relating to the extent to which entities provide valid, fair and transparent arguments to support the decisions made on various critical accounting assumptions and estimates disclosed in the annual reports demonstrates extreme variability, with a coefficient of variation of 28.64%, highlighting dramatic differences between countries in the rigor and transparency of the processes used to develop complex accounting estimates that require the exercise of high quality professional judgment. Estonia emerges as the regional leader in this critical dimension with a score of 2.27, followed by Lithuania with 1.77, while Latvia and Poland demonstrate substantial and worrying deficiencies with scores of 1.09 and 1.33, respectively.
The dimension relating to the extent to which companies base the choice of specific accounting policies on sound, transparent and persuasive theoretical and practical arguments emerges as the dominant component of faithful representation across the sample analyzed, with an impressive mean of 3.76 indicating an overall satisfactory level of concern for the justification of accounting choices, but simultaneously exhibiting considerable variability with a coefficient of variation of 18.82% suggesting substantial differences across jurisdictions in the rigor and completeness of the processes of accounting policy selection and justification. Poland is the absolute leader with an exceptional score of 4.45, closely followed by Romania with 4.20 and Lithuania with 4.20.
The qualitative characteristic of comparability of financial statements, defined as the extent to which accounting information allows users to identify similarities and differences between economic phenomena both within the same entity over time and between different entities in the same period, presents results that fundamentally contradict the proclaimed objectives of international accounting harmonization and reveal the existence of deep structural barriers to achieving effective material harmonization in the Central and Eastern European region. The statistical F-test indicates highly significant differences between countries (F = 5.83, p < 0.001), with differences of up to 33% between the extreme values of the comparability index, demonstrating that the formal adoption of the same set of international financial reporting standards has not been sufficient to eliminate significant fragmentation of accounting practices and to ensure a satisfactory level of uniformity in the application of fundamental conceptual principles that facilitate effective comparability of financial information across jurisdictions and entities.
The results highlight the need to intensify international accounting harmonization efforts. The persistent 33% difference in the extremes of the comparability scores, despite uniform IFRS adoption, provides empirical evidence of the ‘form over substance’ implementation phenomenon identified by Ball et al. (2003).
The dimensions relating to comprehensively and transparently explaining accounting policy changes and providing sufficient detail to facilitate cross-sectoral and cross-national comparability are the main sources of differentiation between countries, with extremely strong correlations with the average comparability score (r > 0.90), indicating that these aspects are the critical elements that decisively determine the ability of users to make relevant and useful comparisons between financial information provided by entities in different jurisdictions in the region.
The comprehensive analysis of the understandability of financial information, defined as the extent to which accounting information is presented in a clear, organized and accessible manner that facilitates understanding and effective use by users with reasonable knowledge of business and economic activities, reveals a surprising and problematic dichotomy between the technical dimensions of information presentation and those specifically oriented to the needs and capabilities of users, with important implications for democratizing access to financial information and for the effectiveness of accounting communication in supporting the decision-making processes of various categories of users of accounting information. This internal fragmentation of performance in the understandability dimension suggests the existence of a predominantly formal and technocratic approach to financial reporting in the Central and Eastern European region, which focuses on compliance with the technical requirements of standards at the expense of effective and accessible communication with real users of financial information.
Unlike previous studies that relied on subjective manual scoring, our automated NLP approach provides objective, replicable measurements of qualitative characteristics. This represents a significant methodological advancement over studies by Kabwe (2023) and Nguyen (2023), which used traditional manual NiCE assessments. Our standardized scoring algorithm eliminates inter-rater reliability concerns and enables large-scale comparative studies previously impossible due to resource constraints.

4. Materials and Methods

The literature reveals a plethora of established methodologies for assessing the quality of financial reporting, as presented in the article containing the literature review of the methods used to measure the quality of financial reporting (Dănescu & Stejerean, 2023).
Accounting-based methods, including proxies for earnings management (Jones model changes, discretionary accruals), have been widely applied (Dechow & Dichev, 2002; Kothari et al., 2005). However, these approaches focus mainly on the quality of earnings rather than the quality of issuers’ corporate discourse, and hence the quality of publicly available financial reports, and may not capture the full spectrum of qualitative characteristics defined by specific conceptual frameworks.
We have chosen to customize the NiCE methodology for the following reasons: detailed coverage, capturing numerous qualitative characteristics defined by the IASB Conceptual Framework; theoretical grounding, ensuring alignment with globally recognized quality criteria; proven reliability through testing in various contexts (Yurisandi & Puspitasari, 2015), but also its universal applicability, having been successfully implemented in diverse economic environments, from developed markets (Western Europe, North America) to emerging economies (South East Asia, Africa).
Traditional applications of the NiCE methodology have relied on manual analysis, creating bottlenecks in large-scale studies and introducing potential subjectivity, resulting in a small number of studies conducted using this methodology, in contrast to the others that are widespread. Our study addresses the identified limitations by implementing an automated assessment framework using natural language processing (NLP) techniques, deeming them more appropriate in this context than dictionary-based methods (Loughran & McDonald, 2011), which offer a straightforward implementation but lack contextual understanding.
In order to achieve the research objective of identifying the application of an international financial reporting framework to ensure uniformity in the quality of accounting information, we comprehensively assessed the quality of annual financial reports disseminated to investors of companies listed on the main stock exchanges in Central and Eastern Europe in 2019–2023. Specifically, we investigate whether there is convergence in the quality of financial reporting in Central and Eastern Europe by applying a methodology using natural language processing (NLP) techniques.
According to the recommendations identified in the works of Li (2010) and Jones and Shoemaker (2011), NLP techniques are suitable for assessing the linguistic structure and coherence of reports, allowing the measurement of the clarity and detail of terms, thus libraries such as NLP: spaCy and NLTK are used to analyze and categorize the text identified in annual financial reports. Techniques include, but are not limited to term extraction, sentiment analysis, and word-frequency analysis to identify linguistic consistency and clarity.
The methodology for assessing the quality of the annual financial statements is structured considering the four measurable quality dimensions, aligned with the principles of the IASB conceptual framework, but also with the scoring system developed by the IASB: relevance, accurate representation, comparability and understandability, and are detailed in Table 19.
Regarding the automated quality assessment of annual financial reports, we propose the implementation of an algorithm in the Pyton 3.13.5 language, structured in three distinct steps, namely (1) text extraction and processing, (2) linguistic analysis and (3) quantification of the results.
The first stage consists mainly in the preparation of the documents, by extracting the text identified in the annual financial reports, saved in PDF, or HTML, and in English, either using the official websites of the sampled countries or through the websites of the stock exchanges. The text is extracted, using specialized libraries such as PyPDF2, docx, BeautifulSoup, and then transformed into text format for the next step.
The second stage involves the use of linguistic analysis, where the previously extracted text is processed using NLP techniques in order to identify certain recurrent linguistic structures relevant to each evaluation criterion. Taking into account the specificity of the evaluated criteria, regular expression matching methods were applied to identify key terms and phrases, word frequency analysis to assess vocabulary and jargon complexity, organizational structure identification and visual element detection through textual references.
The third stage aims to measure and aggregate the results obtained by implementing a standardized scoring system, aligned with the one proposed by NiCE, which operates on an ordinal scale between 1 and 5 for each of the criteria under evaluation. The scoring algorithm is based on the degree to which the characteristics defined as a priori in the conceptual framework of the study, are met, being judged based on the following cardinal values:
  • Unsatisfactory level of fulfillment of the assessed criterion, suggesting the absence of defining characteristics or their poor implementation—Value 1;
  • Partially Satisfactory level, minimal fulfillment of the defining characteristics—Value 2;
  • Medium level of compliance, presence of defining characteristics but some limitations in implementation—Value 3;
  • Good level of fulfillment of the assessed criterion, with almost complete presence of the characteristics and consistent implementation—Value 4;
  • Optimal level of compliance, all defining features are fully present and implemented to a superior standard—Value 5.
There are many gaps in the existing literature regarding the assessment of quality characteristics in the post-communist economies of Central and Eastern Europe. Ball et al. (2003) argue that the formal adoption of international standards does not automatically guarantee an improvement in the quality of reporting, especially in countries with civil-judicial traditions and weak enforcement institutions. At the same time, Soderstrom and Sun (2007) identify country-specific political and economic factors as moderators of the effects of IFRS adoption on the relevance of financial reporting.
Based on the contradictions identified and the lack of specific empirical evidence for the CEE region, and also to achieve the main objective of this research, we propose five research hypotheses (Table 20), set in the context that specific financial reporting practices remain present in each jurisdiction, along with the possibility of choice in selecting accounting policies, which have effects in defining the level of the four qualitative characteristics. However, we recognize the concern of national and international harmonizers towards international harmonization and the elimination of information asymmetry in the quality of disclosures.
The main selection criteria for the countries included in the research sample are geography and common history, facilitating comparability by eliminating the influence of diverging factors such as significant differences in economic systems. Thus, countries belonging to the former Eastern European bloc of centrally planned economies, characterized by similar socio-political background and synchronized economic transitions post-1989, were selected in the sample.
The geographical and historical delineation of the sample is based on theoretical premises, aligned with previous research in the field of post-communist emerging markets. As emphasized by Berglöf and Bolton, the countries of Central and Eastern Europe have followed a unique trajectory of economic and institutional development, marked by the transition from planned to market economies (Berglöf & Bolton, 2002). This common experience creates, as other researchers argue, a natural analytical framework for comparative studies (Roland, 2000).
Subsequently, the other criteria for the selection of the sampled countries are considered:
  • Consistency, the countries were continuously listed on the specific stock exchanges of each country in the sample during the study period;
  • Availability, i.e., the annual financial reports of the countries are available either on the official website or on the website of the Stock Exchanges during the study period;
  • Inclusion in the index, thus only those countries that are included in the main index of the Stock Exchanges of each state were selected, being considered the most representative and relevant in obtaining pertinent conclusions.
Following the application of all the above criteria, the research sample includes 114 companies, geographically distributed in 10 distinct countries, as presented in Table 21:
Considering the period analyzed, i.e., 2019–2023, this research assesses the quality of financial information of 570 annual reports.

5. Conclusions

The study of empirical data collected from 570 annual reports of entities from ten CEE countries reveals differences in the quality of financial reporting, as revealed by the testing of the five established hypotheses.
Regarding the Relevance of financial reporting (Hypothesis 1), descriptive statistical analysis indicates a partial convergence, manifested by constant scores for characteristics R2, R3 and R4. However, characteristic R1 shows variation across countries, with Croatia and Romania recording the highest values and Latvia and Hungary at the opposite pole. These differences, although moderate in magnitude, reflect distinct approaches to relevance in different jurisdictions.
Regarding the Accurate Representation of financial information (Hypothesis 2), the results indicate statistically significant differences between the analyzed countries (F = 7.36, p < 0.001), rejecting the uniformity hypothesis. Dimension F2 (Compliance with standards) shows the strongest correlation with the overall accurate representation score (r = 0.91), suggesting the critical importance of standardization for accurate representation of reporting. Stratifying the countries into three distinct performance groups reveals gaps of about 20% between the extreme values, reflecting significant discrepancies in the accounting culture of the different jurisdictions.
The analysis of the Comparability of financial statements (Hypothesis 3) also reveals statistically significant differences between countries (F = 5.83, p < 0.001). Criteria C4 (Cross-sectoral detail) and C5 (Cross-national comparability) show the strongest correlations with the overall score (r > 0.90), indicating the substantial contribution of these di-measures to the differentiation of countries. The gap of about 33% between the extreme values of the overall score suggests a significant fragmentation of reporting practices, with implications for the efficient functioning of cross-border financial markets.
Regarding the Intelligibility of financial information (Hypothesis 4), the F-test (F = 4.273, p = 0.002) confirms the existence of significant differences between countries. There is a notable discrepancy between the technical dimensions of understandability (clarity, detail) and the user-oriented dimensions (accessibility, adaptability), reflecting a predominantly formalistic approach to financial reporting. The strong negative correlation between clarity of presentation and user-friendliness (r = −0.730) underlines the in-built tension between technical rigor and practical accessibility of financial information.
The analysis of the overall Financial Reporting Quality index (Hypothesis 5) shows a lack of uniformity at both cross-sectional and longitudinal levels. The variance coefficient of 9.02% for the whole period and the gap of up to 33% between the extreme values contradict the fact that the mere existence of an international reference framework is not sufficient to ensure uniform quality of financial reporting, and that a continuous move towards international harmonization is needed, together with consistent efforts to apply international requirements on the quality of disclosures in published financial reports in a consistent manner.
The generalized deterioration of the quality index in 2023 (−7.07% on average) points to a structural vulnerability of financial reporting systems to disruptive factors, possibly associated with recent global economic challenges. The substantial variability in resilience across countries, reflected in the different magnitude of deterioration (between 5.95% and 16.65%), emphasizes the importance of institutional and cultural factors in determining the robustness of accounting systems.
Overall, the results of the research indicate that, in the context of current international accounting practices, significant differences in the quality of financial reporting persist, driven by jurisdiction-specific contextual factors.
The present study has several limitations that need to be taken into account when interpreting the results, and which can be considered as a starting point for future research. First, automating evaluation using NLP techniques, while improving objectivity and scalability, may not capture the subtle nuances of reporting quality that are apparent to experienced human evaluators, particularly in interpreting complex semantic context and authorial intent. Second, the time interval analyzed (2019–2023) may be insufficient to identify long-term trends, and the inclusion of the pandemic period and the geopolitical crisis of 2022–2023 may distort normal patterns of financial reporting quality evolution. Finally, the NiCE methodology, although extensively validated, focuses on qualitative aspects of reporting and may not fully capture all relevant dimensions of financial reporting quality, such as impact on user decisions or relevance to specific capital markets.
The limitations identified in the present study open multiple opportunities for further research in the field of quality measurement of financial reporting. Further research should focus on developing hybrid algorithms that combine the effectiveness of NLP techniques with the expertise of human evaluators to better capture the complex semantic nuances of financial reporting. Extending the time horizon through longitudinal studies over longer periods would allow the identification of long-term trends and the separation of the impact of exogenous factors from the natural evolution of reporting quality. In addition, the development of complementary methodological frameworks integrating additional dimensions such as impact on user decisions and relevance for capital markets could provide a more comprehensive perspective on the quality of financial reporting.

Author Contributions

Study conception and design: R.M.S.; data collection: R.M.S.; analysis and interpretation of results: R.M.S. and T.D.; draft manuscript preparation: T.D. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Ball, R., Robin, A., & Wu, J. S. (2003). Incentives versus standards: Properties of accounting income in four East Asian countries. Journal of Accounting and Economics, 36(1–3), 235–270. [Google Scholar] [CrossRef]
  2. Beest, F. V., Braam, G., & Boelens, S. (2009). Quality of financial reporting: Measuring qualitative characteristics (Nijmegen Center for Economics (NiCE) Working Paper, 09-108). NiCE. [Google Scholar]
  3. Berglöf, E., & Bolton, P. (2002). The great divide and beyond: Financial architecture in transition. Journal of Economic Perspectives, 16(1), 77–100. [Google Scholar] [CrossRef]
  4. Dănescu, T., & Stejerean, R. M. (2022). Companies’ behavior in measuring the quality of financial reports: Pre-and post-pandemic research. Frontiers in Psychology, 13, 1005941. [Google Scholar] [CrossRef] [PubMed]
  5. Dănescu, T., & Stejerean, R. M. (2023). A screening of tools to measure the quality of financial reporting. Acta Marisiensis. Seria Oeconomica, 17(1), 11–28. [Google Scholar] [CrossRef]
  6. Dechow, P. M., & Dichev, I. D. (2002). The quality of accruals and earnings: The role of accrual estimation errors. The Accounting Review, 77, 35–59. [Google Scholar] [CrossRef]
  7. Eliza, E. (2023). Does SAK online enhance the quality of financial reporting? The Indonesian Journal of Accounting Research, 25, 299–318. [Google Scholar] [CrossRef]
  8. International Accounting Standards Board (IASB). (2018). Conceptual framework for financial reporting. IFRS Foundation. [Google Scholar]
  9. Jones, M. J., & Shoemaker, P. A. (2011). Accounting narratives: A review of empirical studies of content and readability. Journal of Accounting Literature, 13, 142–184. [Google Scholar]
  10. Kabwe, M. (2023). Assessment of financial reporting quality in a developing country using nice qualitative characteristics measurement. International Journal of Finance and Accounting, 8, 1–22. [Google Scholar] [CrossRef]
  11. Kothari, S. L., Leone, A. J., & Wasley, C. E. (2005). Performance matched discretionary accrual measures. Journal of Accounting and Economics, 39(1), 163–197. [Google Scholar] [CrossRef]
  12. Li, F. (2010). Textual analysis of corporate disclosures: A survey of the literature. Journal of Accounting Literature, 29, 143–165. [Google Scholar]
  13. Loughran, T., & McDonald, B. (2011). When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks. The Journal of Finance, 66(1), 35–65. [Google Scholar] [CrossRef]
  14. Nguyen, T. P. H. (2023). The financial reporting quality for commercial banks—Empirical evidence in Vietnam. International Journal of Professional Business Review, 8(6), e01421. [Google Scholar] [CrossRef]
  15. Roland, G. (2000). Transition and economics: Politics, markets, and firms. MIT Press. [Google Scholar]
  16. Soderstrom, N. S., & Sun, K. J. (2007). IFRS adoption and accounting quality: A review. European Accounting Review, 16(4), 675–702. [Google Scholar] [CrossRef]
  17. Wahyuni, E. T., Puspitasari, G., & Puspitasari, E. (2020). Has IFRS improved accounting quality in Indonesia? A systematic literature review of 2010–2016. Journal of Accounting and Investment, 1(1), 19–44. [Google Scholar] [CrossRef]
  18. Whittington, G. (2019). Conceptual framework for financial reporting: Foundation of standard setting. Journal of Financial Reporting, 4(2), 21–49. [Google Scholar]
  19. Yurisandi, T., & Puspitasari, E. (2015). Financial reporting quality—before and after IFRS adoption using NiCE qualitative characteristics measurement. Procedia—Social and Behavioral Sciences, 211, 644–652. [Google Scholar] [CrossRef]
Table 1. Descriptive statistics for Relevance.
Table 1. Descriptive statistics for Relevance.
CriteriaAverageMedianStandard
Deviation
Coefficient
of Variation (%)
MinMax
R11.161.110.1210.341.001.50
R21.001.000.000.001.001.00
R31.001.000.000.001.001.00
R41.001.000.000.001.001.00
Average1.041.030.032.881.001.13
Source: Authors’ projection.
Table 2. Average level of relevance by country (2019–2023).
Table 2. Average level of relevance by country (2019–2023).
Country20192020202120222023Average
Croatia1.081.061.081.131.061.08
Romania1.051.071.081.021.021.05
Czech Republic1.061.031.031.031.061.04
Lithuania1.031.031.031.031.031.03
Estonia1.011.031.031.031.021.02
Latvia1.031.031.001.001.001.01
Polonia1.021.021.021.021.021.02
Hungary1.001.041.021.021.001.02
Bulgaria1.001.001.001.131.001.03
Source: Authors’ projection.
Table 3. Average values by size and country (2019–2023)_Accurate Representation.
Table 3. Average values by size and country (2019–2023)_Accurate Representation.
CountryR1R2R3R4Average
Croatia1.331.001.001.001.08
Romania1.191.001.001.001.05
Czech Republic1.151.001.001.001.04
Lithuania1.111.001.001.001.03
Estonia1.091.001.001.001.02
Latvia1.051.001.001.001.01
Polonia1.091.001.001.001.02
Hungary1.061.001.001.001.02
Bulgaria1.101.001.001.001.03
Source: Authors’ projection.
Table 4. Descriptive statistics for accurate representation (average 2019–2023).
Table 4. Descriptive statistics for accurate representation (average 2019–2023).
CriteriaAverageMedianStandard
Deviation
Coefficient
of Variation (%)
MinMax
F11.471.400.4228.641.002.41
F23.763.890.7118.822.294.82
F31.061.000.1211.831.001.44
F41.011.000.065.841.001.29
F51.021.000.054.581.001.18
Average1.671.680.179.981.341.95
Source: Authors’ projection.
Table 5. Average level of accurate representation by countries (2019–2023).
Table 5. Average level of accurate representation by countries (2019–2023).
Country20192020202120222023MediaSt. DevCV (%)
Lithuania1.921.951.711.671.741.800.137.11
Estonia1.851.881.811.821.581.790.126.81
Polonia1.781.821.751.871.641.770.094.90
Romania1.731.761.771.851.691.760.063.30
Croatia1.591.761.811.871.751.760.116.10
Bulgaria1.601.601.761.881.601.690.137.62
Hungary1.401.631.541.591.401.510.106.93
Latvia1.461.341.631.631.261.460.1611.24
Czech Republic1.421.421.441.441.471.440.021.35
Average1.641.681.691.741.571.660.074.06
St. dev0.180.210.120.170.160.15--
CV (%)11.1012.597.289.7910.179.28--
F Test-----7.36--
p-value-----<0.001--
Source: Authors’ projection.
Table 6. Average values by size and countries (2019–2023).
Table 6. Average values by size and countries (2019–2023).
CountryF1F2F3F4F5Average
Lithuania1.774.201.001.001.021.80
Estonia2.273.671.001.001.001.79
Polonia1.334.451.001.001.071.77
Romania1.564.201.051.001.001.76
Croatia1.364.241.101.061.031.76
Bulgaria1.483.961.001.001.001.69
Hungary1.343.161.061.001.001.51
Latvia1.093.231.001.001.001.46
Czech Republic1.272.761.181.001.001.44
CV (%)28.6418.8211.835.844.589.28
Correlation with average0.580.91−0.220.060.211.00
Source: Authors’ projection.
Table 7. Correlation matrix—Accurate representation.
Table 7. Correlation matrix—Accurate representation.
CriteriaF1F2F3F4F5
F11.0000.159−0.063−0.104−0.102
F20.1591.000−0.2760.0540.284
F3−0.063−0.2761.0000.3190.147
F4−0.1040.0540.3191.0000.154
F5−0.1020.2840.1470.1541.000
Source: Authors’ projection.
Table 8. Descriptive statistics—Comparability (average 2019–2023).
Table 8. Descriptive statistics—Comparability (average 2019–2023).
CriteriaAverageMedianStandard
Deviation
Coefficient
of Variation (%)
MinMax
C12.062.070.5627.181.003.18
C23.844.000.369.482.674.00
C33.914.000.194.803.334.00
C42.142.060.5324.771.253.00
C52.662.680.6725.111.483.73
C63.513.310.9928.121.155.00
Average3.023.100.4113.742.163.65
Source: Authors’ projection.
Table 9. Average level of comparability by country (2019–2023).
Table 9. Average level of comparability by country (2019–2023).
Country20192020202120222023MediaSt. DevCV (%)
Romania3.493.523.423.493.303.440.092.64
Lithuania3.183.433.373.473.653.420.175.10
Croatia3.183.343.373.542.923.270.237.05
Polonia3.473.293.063.212.913.190.216.64
Estonia2.972.963.103.172.762.990.155.12
Hungary2.793.083.113.242.873.020.186.04
Czech Republic2.812.762.742.562.692.710.093.42
Bulgaria2.232.232.803.202.432.580.4216.31
Latvia2.582.562.532.442.162.450.176.98
Media2.973.023.053.152.772.990.144.59
St. dec0.420.420.300.370.390.34--
CV (%)14.0413.929.8811.7114.0411.38--
F Test-----5.83--
p-value-----<0.001--
Source: Authors’ projection.
Table 10. Average values by criteria and countries (2019–2023).
Table 10. Average values by criteria and countries (2019–2023).
CountryC1C2C3C4C5C6Average
Lithuania2.073.844.002.733.404.633.44
Estonia2.063.973.972.742.914.873.42
Polonia2.153.893.942.333.144.163.27
Romania2.823.893.892.712.852.963.19
Croatia2.134.004.002.152.842.842.99
Bulgaria2.493.963.961.972.673.063.02
Hungary1.803.133.871.842.403.222.71
Latvia1.323.523.521.642.083.402.58
Czech Republic1.313.933.931.431.822.312.45
CV (%)27.189.484.8024.7725.1128.1211.38
Correlation with average0.750.290.280.920.930.521.00
Source: Authors’ projection.
Table 11. Correlation matrix—Comparability.
Table 11. Correlation matrix—Comparability.
CriteriaC1C2C3C4C5C6
C11.0000.3270.3270.5610.643−0.079
C20.3271.0000.8010.1050.2870.088
C30.3270.8011.0000.1340.293−0.005
C40.5610.1050.1341.0000.6850.663
C50.6430.2870.2930.6851.0000.518
C6−0.0790.088−0.0050.6630.5181.000
Source: Authors’ projection.
Table 12. Descriptives statistics—Intelligibility.
Table 12. Descriptives statistics—Intelligibility.
CriteriaAverageMedianStandard DeviationMinMaxCoefficient
of Variation (%)
U11.2471.2670.1421.0001.57111.39
U24.2384.2720.6442.6005.00015.19
U33.5433.6610.8761.6424.86724.73
U41.2011.1980.2581.0001.88921.48
U51.1981.2230.1861.0001.66715.53
Average2.2852.3310.3181.6552.62813.91
Source: Authors’ projection.
Table 13. Average level of intelligibility by country (2019–2023).
Table 13. Average level of intelligibility by country (2019–2023).
Country20192020202120222023MediaSt. DevCV (%)
Polonia2.5452.6002.5092.6002.4002.5310.0843.32
Romania2.6132.5332.4402.5472.3732.5010.0933.72
Croatia2.4292.4712.5002.5292.4812.4820.0391.55
Lithuania2.3792.3782.3792.4842.4532.4150.0502.07
Hungary2.2862.4292.6292.6292.2002.4340.1897.75
Estonia2.1412.1532.1532.1061.9882.1080.0673.17
Czech Republic2.2672.0441.9781.9562.4002.1290.1878.80
Bulgaria1.8801.6802.2402.3202.0802.0400.25912.70
Latvia1.6501.9752.0471.9811.7731.8850.1678.86
Average2.2432.2512.3192.3502.2392.2810.0472.05
St. dev0.3100.3100.2180.2610.2300.237--
CV (%)13.8313.789.4011.1110.2810.39--
F Test-----4.273--
p-value-----0.002--
Source: Authors’ projection.
Table 14. Average values by size and country (2019–2023).
Table 14. Average values by size and country (2019–2023).
CountryU1U2U3U4U5Average
Polonia1.2914.3454.3451.2001.4732.531
Romania1.3074.3604.5731.0001.2672.501
Croatia1.2014.8293.9451.1141.3202.482
Lithuania1.2654.4464.1471.0441.1702.415
Hungary1.4004.7713.8141.1291.0572.434
Estonia1.3763.9183.2121.0351.0002.108
Czech Republic1.3113.3113.1111.5561.3562.129
Bulgaria1.0003.8802.8401.4801.0002.040
Latvia1.0294.2881.9031.2061.0001.885
CV (%)11.3915.1924.7321.4815.5310.39
Correlation with average0.2510.3970.927−0.2570.3121.000
Source: Authors’ projection.
Table 15. Correlation matrix—Intelligibility.
Table 15. Correlation matrix—Intelligibility.
CriteriaU1U2U3U4U5
U11.0000.2150.547−0.295−0.130
U20.2151.0000.382−0.730−0.159
U30.5470.3821.000−0.5520.218
U4−0.295−0.730−0.5521.0000.068
U5−0.130−0.1590.2180.0681.000
Source: Authors’ projection.
Table 16. Evolution of the financial reporting quality index (2019–2023).
Table 16. Evolution of the financial reporting quality index (2019–2023).
Country20192020202120222023MediaSt. DevCV (%)
Croatia8.278.638.779.068.218.590.354.08
Polonia8.828.738.348.717.978.510.354.10
Hungary7.478.188.308.477.477.980.475.92
Czech Republic7.567.257.196.987.617.320.263.51
Romania8.898.888.728.918.388.760.222.52
Estonia7.978.038.098.127.347.910.324.07
Lithuania8.508.798.488.668.878.660.171.96
Latvia6.726.917.207.056.196.810.395.68
Bulgaria6.716.517.808.537.117.330.8311.31
Media7.887.998.108.287.687.990.222.70
St. dev0.860.920.620.800.820.72--
CV (%)10.8911.497.699.6910.689.02--
Min6.716.517.196.986.196.81--
Max8.898.888.779.068.878.76--
Amplitude2.182.371.582.082.681.95--
Source: Authors’ projection.
Table 17. Annual rates of change in quality index (%).
Table 17. Annual rates of change in quality index (%).
Country2020/20192021/20202022/20212023/2022CAGR (2019–2023)
Croatia4.351.623.31−9.38−0.18
Polonia−1.02−4.474.44−8.50−2.50
Hungary9.501.472.05−11.810.00
Czech Republic−4.10−0.83−2.929.030.16
Romania−0.11−1.802.18−5.95−1.47
Estonia0.750.750.37−9.61−2.06
Lithuania3.41−3.532.122.421.07
Latvia2.834.20−2.08−12.20−2.04
Bulgaria−2.9819.829.36−16.651.46
Average1.401.912.09−7.07−0.62
Source: Authors’ projection.
Table 18. Matrix of correlation coefficients between quality index values (2019–2023).
Table 18. Matrix of correlation coefficients between quality index values (2019–2023).
CountryCroatiaPolandHungaryCzech
Republic
RomaniaEstoniaLithuaniaLatviaBulgaria
Croatia1.000.720.89−0.930.910.940.050.470.59
Polonia0.721.000.63−0.480.580.68−0.460.180.11
Hungary0.890.631.00−0.840.680.83−0.070.370.57
Czech
Republic
−0.93−0.48−0.841.00−0.88−0.890.09−0.63−0.40
Romania0.910.580.68−0.881.000.890.210.650.44
Estonia0.940.680.83−0.890.891.00−0.050.650.44
Lithuania0.05−0.46−0.070.090.21−0.051.000.130.19
Latvia0.470.180.37−0.630.650.650.131.000.41
Bulgaria0.590.110.57−0.400.440.440.190.411.00
Source: Authors’ projection.
Table 19. Quality characteristics assessed.
Table 19. Quality characteristics assessed.
Quality
Characteristics
CriteriaResearch Question Pursued
RelevanceR1To what extent does the presence of the outlook statement contribute to the formation of expectations and forecasts about the future of the entity?
R2To what extent does the presence of non-financial information complement financial information?
R3To what extent does the company use fair value rather than historical cost?
R4To what extent do the results presented provide feedback to users of annual reports on how various market events and significant transactions have affected the entity?
Accurate
representation
F1To what extent are valid arguments provided to support the decision for certain assumptions and estimates in the annual report?
F2To what extent does the company base its choice of accounting policies on valid arguments?
F3To what extent does the company highlight both positive and negative events in the discussion of annual results?
F4What type of audit opinion is included in the annual report?
F5To what extent does the company provide information on corporate governance
ComparabilityC1To what extent do the explanatory notes on changes in accounting policies explain information about the change in accounting policies?
C2To what extent do the explanatory notes on revisions to estimates explain information on changes in estimates?
C3To what extent has the company adjusted prior period figures for the effect of implementing a change in accounting policy or revisions to accounting estimates?
C4To what extent does the entity provide a comparison of the current accounting period with the previous accounting period?
C5To what extent is the information in the annual report comparable with the information provided by other organizations?
C6To what extent does the company present financial figures and indicators in its annual reports?
IntelligibilityU1To what extent is the annual report presented in a well-organized manner?
U2To what extent are the explanatory notes presented clearly enough?
U3To what extent does the presence of graphs and tables clarify the information presented?
U4How easy is it to follow the use of language and technical reasoning?
U5What is the size of the glossary?
Source: Authors’ projection.
Table 20. Research hypothesis.
Table 20. Research hypothesis.
HypothesisDescriptionQuality
Characteristics
1There is no significant convergence of results on the Relevance of reported financial information between the sample countries.Relevance
2There is no significant convergence of results across sample countries in terms of Accurate Representation of Reported Financial Information.Accurate
representation
3The structure and content of the financial reports published in the selected countries do not ensure an adequate level of comparability.Comparability
4The financial reports published in the selected companies and countries do not ensure an adequate level of Intelligibility in the presentation of financial information.Intelligibility
5At global level, the financial reports do not demonstrate a uniform level of Quality.Global quality
Source: Authors’ projection.
Table 21. Research sample.
Table 21. Research sample.
CountryStock ExchangeMain IndexNo. Companies
CroatiaZagreb Stock ExchangeCROBEX14
RomaniaBucharest Stock ExchangeBET2015
PolandWarsaw Stock ExchangeWIG2011
HungaryBudapest Stock ExchangeBUX14
Czech RepublicPrague Stock ExchangePX9
BulgariaNational Stock Exchange of BulgariaSOFIX5
EstoniaNasdaq Baltic StockOMX Tallinn GI17
LatviaLatvia Stock Market IndexRiga8
LithuaniaLithuania Stock Market IndexVilnius14
SerbiaSerbia Stock MarketBELEX157
Total 114
Source: Authors’ projection.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Dănescu, T.; Stejerean, R.M. An X-Ray Using NLP Techniques of Financial Reporting Quality in Central and Eastern European Countries. Int. J. Financial Stud. 2025, 13, 142. https://doi.org/10.3390/ijfs13030142

AMA Style

Dănescu T, Stejerean RM. An X-Ray Using NLP Techniques of Financial Reporting Quality in Central and Eastern European Countries. International Journal of Financial Studies. 2025; 13(3):142. https://doi.org/10.3390/ijfs13030142

Chicago/Turabian Style

Dănescu, Tatiana, and Roxana Maria Stejerean. 2025. "An X-Ray Using NLP Techniques of Financial Reporting Quality in Central and Eastern European Countries" International Journal of Financial Studies 13, no. 3: 142. https://doi.org/10.3390/ijfs13030142

APA Style

Dănescu, T., & Stejerean, R. M. (2025). An X-Ray Using NLP Techniques of Financial Reporting Quality in Central and Eastern European Countries. International Journal of Financial Studies, 13(3), 142. https://doi.org/10.3390/ijfs13030142

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Article metric data becomes available approximately 24 hours after publication online.
Back to TopTop