An Innovative Model for Assessing Intellectual Capital Based on Information from Corporate Reporting and ESG Factors

Abstract

This paper analyzes the measurement of intellectual capital in the context of the increasing importance of ESG factors and current economic changes, highlighting the role of intangible assets in supporting company performance. The existing literature emphasizes the limitations of traditional models, such as the Value-Added Intellectual Coefficient (VAIC), which do not adequately capture the contribution of the modern dimensions of intellectual capital. The study is based on a quantitative approach and uses a sample of 75 companies listed on the Bucharest Stock Exchange. The data were analyzed using IBM SPSS Statistics, Version 26.0 through the application of principal component analysis (PCA), linear regression, and ANOVA tests. The results show that the traditional model does not significantly explain market performance measured by Tobin’s Q, while the modern model, based on human, structural, relational capital, and ESG factors, provides a more comprehensive conceptual perspective, although its statistical explanatory power remains limited. The paper contributes to the existing literature by proposing an extended approach to intellectual capital evaluation, adapted to the current context, and offers useful implications for investors, managers, and other users of financial information.

1. Introduction

The concept and implications of intellectual capital are currently being addressed with great attention by politicians, investors, small and large companies, banking institutions, and current and future professionals in all fields (Zaytsev et al., 2020). There are optimistic predictions regarding its appreciation and recognition. Its contribution is important for achieving performance, improving economic activity, and social life. Pressure and expectations are increasing because discussions about intellectual capital (IC) have become more than a trend. From another point of view, there is a growing expectation for the implementation of clear standards for measuring and reporting it, which should be practical and accessible to various types of organizations (Andriessen, 2004; Matos et al., 2020). Specialists in the field are concerned about how companies will adopt assessment models so that their values are easy to interpret and compare.

Even though intellectual capital is recognized and valued, its assessment still has shortcomings. One problem is related to identifying clear, comparable market criteria and determining how they can be integrated into organizations’ financial statements (Zaytsev et al., 2020). The traditional VAIC model does not cover all elements of intellectual capital, and the relationship between IC and company performance is not fully explained (Ahmad, 2023). The values obtained through the traditional model do not fully reveal the relationship between intellectual capital and its three components—human, structural, and relational—and their impact on performance.

Moreover, studies have not validated the hypothesis that the traditional model illustrates the value of intellectual capital, establishing that VAIC does not have sufficient connections with the market value of the organization (Ståhle et al., 2011). Some studies emphasize that operational efficiency, as part of the intellectual capital coefficient, does not actually include specific IC elements, relying solely on figures obtained from companies’ annual reports, as the cost of human resources is not sufficient to reflect the contribution of IC (Bakhsha et al., 2017). On the other hand, we must acknowledge that it is very difficult to create a perfect model that captures all aspects considered important, depending on the user’s position. Awareness of the role of IC and emphasis on the need to measure the effectiveness of value creation have attracted increased interest, as evidenced by numerous discussions and research based on VAIC (Pulic, 2000). There are also numerous studies that have obtained important results, such as the positive relationship between IC and financial performance indicators: productivity, profitability, or efficiency (Bansal et al., 2024).

More and more authors are trying to analyze traditional methods or propose new, improved ones, so that they also cover the evaluation of intangibles and sustainability factors (Andriessen, 2004; Frutos-Belizón et al., 2019). In the context of increasingly widespread access to technology, social capital and professional skills can be used as pillars of digital transformation, further expanding the conceptual boundaries of intellectual capital (Švarc et al., 2021). The extended VAIC models have led to the establishment of new components, such as spiritual capital (Hanif et al., 2019) or innovation capital in the banking sector (Buenaño et al., 2025). There is research that finds, using extended VAIC, that the greatest influence on IC is the dimension of human capital (Evangelista et al., 2025). However, the integration of ESG factors as a structural dimension of intellectual capital remains insufficiently explored in the literature.

When it comes to measuring market performance, Tobin’s Q method also has limitations: it does not calculate company performance directly, because underinvestment grows much more than Tobin’s Q drops (Dybvig & Warachka, 2015), and it is not suitable for comparing companies in different fields or of different sizes (Zaytsev et al., 2020). The main purpose of this method was not to measure intellectual capital, as its usefulness is greater when entities are compared with each other over a longer period of time (Valizadeh Morady, 2013). Other authors argue that an analysis of the Q ratio at different points in time can result in an average rate of change (Van den Berg, 2003). The method of calculating the difference between market value and book value has limitations due to large fluctuations in share prices, a situation that can negatively influence the measurement of intellectual capital (Nawrocki, 2022).

The balanced scorecard model also presents limitations, including the undervaluation of human resources, reduced attention to intellectual capital components, and the difficulty of making comparisons across different industries (Hudson et al., 2001). These limitations, considered collectively, reinforce the need for a more comprehensive and adaptable evaluation framework.

To the best of the authors’ knowledge, empirical studies that integrate ESG factors into IC evaluation on the Romanian capital market remain limited. This gap—both methodological and contextual—justifies the need for the present study. The present study integrates ESG factors as a structural dimension of intellectual capital—not merely as a control variable—and empirically tests this extended model on the Bucharest Stock Exchange (BVB). Through this dual contribution, the study addresses both a methodological and a contextual gap identified in the literature.

The central question this study seeks to answer is to what extent the integration of sustainability factors and non-financial dimensions can contribute to improving existing models and overcoming the limitations of traditional methods of intellectual capital assessment. To answer this question, the study proposes an extended model integrating four dimensions: human capital (CU), structural capital (CS), relational capital (CR), and ESG factors, tested on a sample of 75 companies listed on the Bucharest Stock Exchange (BVB), for the period 2018–2023.

The central hypothesis (H0) of the study is formulated as follows: the inclusion of non-financial indicators—human capital, structural capital, relational capital, and ESG factors—in the evaluation of intellectual capital does not significantly improve the explanatory power compared to the traditional VAIC model.

H1 is formulated as follows: traditional models such as VAIC do not clearly capture the contribution of intangible assets associated with the non-financial dimensions of company performance. Compared to market performance measured by Tobin’s Q, the explanatory power of the traditional model is statistically insignificant.

H2 to be tested states that a modern intellectual capital evaluation model that includes human, structural, relational capital, and ESG factors has broader conceptual coverage in relation to the market value of listed companies, compared to traditional models that do not incorporate the sustainability dimension.

The article is structured as follows: Section 2 presents the methodological framework and the data used. Section 3 presents the results obtained through the two models—traditional and modern. Section 4 discusses the results in relation to the existing literature, and Section 5 presents the conclusions of the study, its limitations, and directions for future research.

2. Materials and Methods

The research strategy adopted in this study is quantitative, based on statistical and econometric methods. The analysis focuses on the evaluation of intellectual capital using both a traditional approach and a modern approach that integrates the four dimensions: human capital (CU), structural capital (CS), relational capital (CR), and ESG factors. The quantitative perspective is based on financial and non-financial data from a sample of 75 listed companies. The main objective is to test the proposed hypotheses and to analyze the influence of intellectual capital determinants on economic performance. The methodology integrates principal component analysis (PCA) to construct a composite intellectual capital score, along with linear regression and ANOVA to examine the relationship between intellectual capital and performance. The literature review provides the theoretical foundation for the development of the research hypotheses. The data submitted for analysis was selected from the financial statements and governance reports on the official website of the Bucharest Stock Exchange, as well as from the official websites of companies or public institutions such as the National Agency for Public Finance (ANAF).

The proposed evaluation model is not entirely new, but it extends existing intellectual capital (IC) assessment methodologies by incorporating ESG factors. Within this study, ESG factors are interpreted as an extended dimension that influences intellectual capital and may contribute to the creation of intangible value.

The database constructed for this analysis includes 75 companies listed on the Bucharest Stock Exchange at the beginning of the analyzed period, 2018. The selection included companies from all sectors of activity available on the Romanian capital market, namely: professional, scientific and technical activities, agriculture, forestry and fishing, wholesale and retail trade, construction, hotels and restaurants, real estate rental, extractive industry, information and communications, and transport and storage. The main inclusion criterion was the availability of public financial data for at least one year of the period 2018–2023.

Of the 75 companies initially included, 6 were delisted or entered bankruptcy or insolvency during the analyzed period: SC Foraj Sonde SA (delisted in 2024), Comtram SA (withdrawn from trading), SC Construcții Montaj SA (entered bankruptcy), ICMRS SA (delisted in 2021), Athenee Palace SA (delisted in January 2021), and SC Prahova Estival 2002 SA (delisted in January 2022, insolvency). For these companies, data are available only for the years prior to delisting, which explains the difference between the theoretical maximum number of observations (75 companies × 6 years = 450) and the number actually used in the analysis (426 observations).

According to Figure 1, the analysis will be carried out taking into account the data of the 75 companies, divided into 9 areas of activity.

Establishing these assumptions and addressing this topic is justified by the changes facing society as a whole: the need to limit risk, the need to automate repetitive and mechanical tasks, the efficient use of employees’ working time, creating future strategies that increase performance while streamlining resources, analyzing large data sets as accurately as possible using algorithms, and the increasingly frequent use of technology. Company management is increasingly focused on generating new ideas and leveraging knowledge so that the greatest effort is connected to creating added value, rather than performing repetitive and mechanical tasks.

The analysis and processing of numerical data was performed using IBM SPSS Statistics software, and the results were subsequently interpreted, and the factors that most influenced the results were identified.

We used linear regression to test the replicable model for calculating intellectual capital. Traditional valuation methods focus primarily on financial indicators such as profitability or book value. The shift to modern methods is essential in order to include non-financial factors that contribute to competitive advantages. A comprehensive intellectual capital concept integrates both ESG indicators and information on organizational processes, relationships with suppliers, customers, banking institutions, reputation, and policies to encourage innovation.

To highlight the contribution of intellectual capital to the company’s value creation, we calculated intellectual capital using two models: traditional and multidimensional (CU, CS, CR, ESG). We extracted the variables presented in

Table 1. Selection of variables.

No.	Nature of Variable	Description	Acronym
1.	Main/endogenous dependence	Measures company performance	TQ
2.	Alternative/endogenous dependence	Asset efficiency	ROA
3.	Traditional/exogenous independence	Traditional model for measuring IC	VAIC
4.	Integrated model/exogenous independence	Human capital	CU
5.	Integrated model/exogenous independence	Structural capital	CS
6.	Integrated model/exogenous independence	Relational capital	CR
7.	Integrated model/exogenous independence	ESG factors	ESG

to build the models.

To prove the hypotheses, we decided to test them using the SPSS program, as shown in

Table 2. Model and selected variables.

No.	The Proposed Mathematical Model	Dependent Variables	Independent Variables	Formula Applied	Purpose of the Model
1.	IC replication model	IC	CU, CS, CR, ESG	Linear regression OLS: IC = α + β1CU + β2CS + β3CR + β4ESG + ε	R2, p-value, β coefficients
2.	IC_PCA model	IC_PCAcomposite score	VAIC	Principal Component Analysis (PCA)	Factor loadings, % variance explained
3.	Traditional firm performance model	Tobin’s Q	VAIC	Linear regression OLS: TQ = α + βVAIC + ε	R2, significance coefficient
4.	Modern company performance model	Tobin’s Q	CU, CS, CR, ESG	Multiple regression: TQ = α + β1CU + β2CS + β3CR + β4ESG + ε	R2, ΔR2 vs. traditional model
5.	Exploratory robustness analysis (MLP)	Tobin’s Q	CU, CS, CR, ESG	Multilayer Perceptron (MLP)—SPSS	R2, RMSE, comparison with linear regression

.

Using linear regression, we calculated the value of intellectual capital based on simple indicators, using the following formula:

IC = α + β₁CU + β₂CS + β₃CR + β₄ESG

The following section presents the results obtained by applying this methodology to the sample of 75 companies listed on the Bucharest Stock Exchange.

3. Results

The results obtained from the analysis of the 75 companies are presented separately for each model. The exploratory robustness analysis is performed using the Multilayer Perceptron (MLP) method. Each stage makes a considerable contribution to a better understanding of how the dimensions of intellectual capital influence company performance.

3.1. Modern IC Model

From our sample, we identified several simple indicators that facilitate comparability across companies (

Table 3. Simple indicators used for the PCA model (modern model).

Communalities	Initial
Return on Assets	1.000
Solvency Ratio	1.000
PER	1.000
Book Value per Share	1.000
Gross Productivity	1.000
Net Profit per Employee	1.000
Market Capitalization/Equity	1.000

). This enabled the construction of a composite intellectual capital (IC) score through principal component analysis (PCA). The aggregated scores represent the predictors used in our model for the three dimensions of intellectual capital: human capital (CU), relational capital (CR), and structural capital (CS). These scores resulted from reducing the data volume of seven financial and non-financial indicators. If raw values from the financial statements had been used, there would have been a risk that indicators with very large values would dominate the analysis, even if they were not the most relevant. In addition, the interpretation of the model would have become more difficult, while its robustness would have been reduced.

One of the important methodological aspects of this study is the way in which the ESG indicators were constructed. Standardized ESG data are not systematically available for companies listed on the Bucharest Stock Exchange during the analyzed period, 2018–2023, which is a challenge frequently encountered in research on emerging markets (Farooq, 2015; Karyani & Perdiansyah, 2022). For this reason, we opted for the construction of proxy variables derived from publicly available financial statements and governance reports.

The environmental dimension (E) was calculated as the ratio of net investment cash flow to total assets. We consider that investment intensity indirectly reflects the company’s commitment to long-term development and responsible environmental practices. The social dimension (S) was measured through net profit per employee, an indicator that captures the value created per worker and is consistent with approaches in the intellectual capital literature regarding human capital efficiency (Nadeem et al., 2019). The governance dimension (G) was represented by the general solvency ratio, as the literature confirms a positive relationship between ESG performance and corporate solvency, with financially stable companies tending to exhibit stronger governance practices (Brogi et al., 2022; Yu & Su, 2024).

The composite ESG score was calculated as the arithmetic mean of the three normalized sub-scores, in order to ensure a balanced contribution of each dimension: ESG = (E_normalized + S_normalized + G_normalized)/3.

To ensure a balanced representation of each variable’s contribution through the composite score, the selected indicators were standardized prior to applying principal component analysis (PCA). The price-to-earnings ratio (PER) is the ratio of market price to earnings per share, illustrating how the market is valued based on profitability. In contrast, Tobin’s Q (TQ), used in this study as the dependent variable, captures the relationship between market value and book value of assets, serving as an indicator of market performance.

Thus, the PER is used only as a complementary indicator to support the interpretation of market perception, while the core empirical analysis is based on TQ. Using SPSS, we focused on the simple indicators presented in

Table 4. Total variance explained (modern model).

Component	Initial Eigenvalues			Loadings			Rotation Sums of Squared Loadings
	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	1.950	27.855	27.855	1.950	27.855	27.855	1.940	27.716	27.716
2	1.133	16.187	44.043	1.133	16.187	44.043	1.136	16.222	43.938
3	1.009	14.420	58.463	1.009	14.420	58.463	1.012	14.458	58.396
4	1.003	14.323	72.785	1.003	14.323	72.785	1.007	14.390	72.785
5	0.922	13.170	85.956
6	0.914	13.053	99.009
7	0.069	0.991	100.000

Extraction method: principal component analysis.

, derived from the financial data of companies listed on the Bucharest Stock Exchange, to construct a composite intellectual capital score through dimensionality reduction. A standardized score was used to ensure comparability over the 2018–2023 period.

We note that the first four components have values greater than one, and if we add them together, we get 72.785%, which means that the degree of explanation in the total variance is significant. This percentage is appropriate to serve as a premise in the process of establishing a composite IC score.

The rotated component matrix illustrated in

Table 5. Rotated component matrix (modern model).

	Component
	1	2	3	4
Return on Assets (ROA)	0.030	−0.049	0.944	−0.047
Solvency Ratio	0.007	−0.011	−0.035	0.964
PER	−0.071	0.664	−0.066	0.034
Book Value per Share	0.117	0.600	−0.120	−0.194
Gross Productivity	0.980	0.016	0.017	0.000
Net Profit per Employee	0.979	0.021	0.022	−0.002
Market Capitalization/Equity	−0.014	−0.576	−0.315	−0.191

Extraction method: principal component analysis. Rotation method: Varimax with Kaiser normalization. Rotation converged in 5 iterations.

provides evidence that the theoretical model functions as intended, namely that the four dimensions of intellectual capital (IC) are integrated into the model.

The Kaiser criterion identified four components consistent with the proposed model, each with eigenvalues greater than 1. Within Component 1, the highest loadings are associated with gross productivity (0.980) and net profit per employee (0.979). For our sample, these values establish human capital as the main differentiating factor and also illustrate the efficiency of human capital.

Consistency with relational capital is captured in Component 2, where the highest loadings are associated with the Price-to-Earnings Ratio (PER), at 0.664, and book value per share, at 0.600. These results demonstrate that the model incorporates information related to investor perception and the dimension associated with external relationships within intellectual capital.

The efficiency of structural capital is captured by return on assets within Component 3, with a loading of 0.944. The high value of this indicator suggests that companies use their available organizational resources efficiently.

The financial stability of companies and the existence of governance practices are reflected in the solvency ratio of 0.964 within Component 4. Companies recording high values for this indicator are largely self-financed and may be considered financially stable. Furthermore, we consider that these companies exhibit corporate governance practices and effectively integrate ESG factors.

The graphical representation in Figure 2 below indicates that, if the elbow method is applied, the components with the highest values—1.950 and 1.133—could be retained. On the other hand, the Kaiser criterion identifies four components with eigenvalues greater than 1, which explain the variance.

The statistical results obtained for the four components demonstrate that the proposed conceptual model integrates and retains all four dimensions.

After processing the database for selected companies listed on the Bucharest Stock Exchange, we found that 99.9% of the variation in intellectual capital is explained by four variables: human capital, structural capital, relational capital, and sustainability indicators (ESG),

Table 6. Model summary for IC calculation (modern model).

Model	R	R-Square	Adjusted R-Square	Std. Error of the Estimate	Change Statistics
					R-Square Change	F Change	df1	df2	Sig. F Change (p)
1	0.999 a	0.999	0.999	9223.372036854777000	0.999	73,734.013	4	421	0.000

^a Predictors: (Constant), ESG, CS, CR, CU.

.

The very high value of the coefficient of determination should be interpreted with caution. The composite intellectual capital score is built from the same conceptual dimensions—human capital (CU), structural capital (CS), relational capital (CR), and ESG factors—that are subsequently used as explanatory variables in the regression model. Under these conditions, the high explanatory power largely reflects the structural relationship between the component variables and the resulting aggregate indicator, rather than representing an independent validation of the model. The near equality between R² and adjusted R² suggests strong internal consistency. The value of 73,734.013 for F Change and 0 for Sig. F. Change demonstrates that, from a statistical point of view, this model is highly significant. In this context, the value of intellectual capital, influenced by CU, CS, CR, and ESG, is much more significant and complex, and its variation is much easier to understand. The model created based on the four dimensions,

Table 7. ANOVA test results (modern model).

Model		Sum of Squares	df	Mean Square	F	Sig.
1	Regression	179,036,059,059,341.700	4	44,759,014,764,835.420	73,734.013	0.000 a
	Residual	255,561,097,670.185	421	607,033,486.152
	Total	179,291,620,157,011.880	425

^a Dependent Variable: IC. Predictors: (Constant), ESG, CS, CR, CU.

, is relevant, and the F test, where F = 73,734.013, p < 0.001, demonstrates that it is based on a valid methodology and reasoning.

The ANOVA test results demonstrate that the model proposed in our research, based on the four dimensions, is globally significant, with an F value of 73,734 and a p-value of less than 0.001. Additionally, the results show that the dimension that most influences IC is human capital, with 0.987. Its value is statistically significant (p is less than 0.001), thus becoming the main determinant of intellectual capital. If we analyze the Unstandardized coefficient, we find that an increase in HC by a single unit will have the effect of increasing IC by 0.329 units (

Table 8. Linear regression results: impact of human, structural, relational, and ESG capital on market performance (modern model).

Model		Unstandardized Coefficients		Standardized Coefficients	t	Sig. (p-Value)	Collinearity Statistics
		B	Std. Error	Beta			Tolerance	VIF
1	(Constant)	−1084.695	1253.373		−0.865	0.387
	CU	0.329	0.006	0.987	52.717	0.000	0.010	103.584
	CS	0.349	0.265	0.002	1.316	0.189	0.998	1.002
	CR	15.394	48.997	0.001	0.314	0.754	0.991	1.009
	ESG	0.024	0.038	0.012	0.645	0.519	0.010	103.499

). According to Collinearity Statistics, a high collinearity was established between the HC and ESG dimensions.

Looking at the results in Table 8, we can observe that human capital has the strongest influence on the composite IC score (β = 0.987; p < 0.001). An increase of one unit in human capital leads to an increase of 0.329 units in the intellectual capital score. We note that this result should be read carefully. The composite IC score is built from the same dimensions that are used as predictors in the regression, which means that part of this explanatory power reflects the structural relationship between the variables rather than an independent confirmation of the model. Additionally, the high multicollinearity between CU and ESG (VIF > 100) limits the reliability of individual coefficients. With these considerations in mind, we treat human capital as the most prominent dimension in our model, while acknowledging that the strength of this relationship is partly a consequence of the way the model was constructed.

The high VIF values for the human capital (103.584) and ESG (103.499) dimensions suggest a conceptual and methodological overlap between these two components. This situation may affect the stability of the coefficients in the combined model. To clarify the independent contribution of ESG factors, we ran a separate regression analysis. The independent variable is the ESG score alone, while the dependent variable is market performance measured by Tobin’s Q.

The results of this analysis confirm that ESG, analyzed independently, does not significantly explain the variation in market performance (R² = 0.000; F = 0.162; p = 0.687). The standardized coefficient Beta = −0.019 is statistically insignificant (p = 0.687), and the VIF value of 1.000 confirms the absence of multicollinearity when ESG is analyzed separately from the other components.

Although we obtained high VIF values, above 100, for human capital (VIF = 103.584) and ESG (VIF = 103.499), the arguments for retaining the integrated model involve two essential aspects.

The first argument relates to the main objective of the model—assessing the aggregate explanatory power of the four intellectual capital dimensions on market performance, rather than estimating individual coefficients precisely. Multicollinearity limits the interpretation of individual coefficients, but does not affect the overall quality of the model. It is important to note that the model remains globally significant (F = 73,734.013, p < 0.001). Additionally, CS and CR present VIF values of 1.002 and 1.009, respectively. This shows that multicollinearity occurs strictly between CU and ESG, without affecting the other dimensions of the model.

The second argument relates to the conceptual overlap between CU and ESG. This reflects an empirical reality specific to emerging markets, where the proxy construction of ESG indicators inevitably shares variance with human capital indicators. Consequently, the interpretation of the combined model is conducted at the conceptual level, rather than through the lens of individual coefficients, while the distinct effect of ESG is assessed through a separate regression.

These results are consistent with the H0 hypothesis of the study. The analysis highlights that the identified limitation is not methodological in nature but rather reflects an empirical reality: in the context of the Romanian capital market, ESG factors constructed on the basis of publicly available data do not have significant independent explanatory power over the market value of listed companies. This conclusion opens important directions for future research, particularly regarding the availability and standardization of ESG data in emerging markets.

3.2. Traditional Performance Model

The regression analysis shows that the traditional VAIC model is not particularly relevant in interpreting the market performance of the listed companies included in our analysis. The positive weights of both structural capital and sustainability indicate a favorable effect on performance and IC value.

The value of the coefficient of determination in

Table 9. Results following the selection of the VAIC variable (traditional model).

Model Summary
R-Square	Adjusted R-Square	Std. Error of the Estimate
0.000	−0.002	4.198492290607874

Predictors: (Constant), VAIC MOD.

(R² = 0.000) demonstrates that the traditional model does not provide the user with information regarding the variation in performance calculated using Tobin’s Q, and the negative value of the Adjusted R-Square indicates limited explanatory power. Determinants such as sustainability and digital transformation could offer much greater explanatory power and accuracy.

The results suggest that the model has considerable limitations regarding the presentation of the connections between market performance and intellectual capital (

Table 10. Results after selecting the dependent variable TQ, according to ANOVAa (traditional model).

Model		Sum of Squares	df	Mean Square	F	Sig.
1	Regression	0.182	1	0.182	0.010	0.919 a
	Residual	7456.364	423	17.627
	Total	7456.546	424

^a Dependent Variable: TQ. Predictors: (Constant), VAIC MOD.

).

Following the regression analysis,

Table 11. Estimation of the linear regression equation, according to Coefficientsa (traditional model).

Model		Unstandardized Coefficients		Standardized Coefficients	t	Sig.
		B	Std. Error	Beta
1	(Constant)	1.092	0.204		5.359	0.000
	VAIC MOD	1.872 × 10−5	0.000	0.005	0.102	0.919

Dependent Variable: TQ.

shows that the traditional VAIC model is not particularly relevant in interpreting the market performance of the listed companies included in our analysis. The coefficient of determination is almost zero, and the link with the Tobin coefficient is of little significance.

The regression coefficient specific to the VAIC MOD variable is very small, 1.872 × 10⁻⁵, and the Beta coefficient with a value of 0.005 demonstrates a low connection between IC measured by VAIC and the market performance of companies. The t-test value of 0.102 is an indication that the traditional model is unable to illustrate the true contribution of IC to performance. The traditional model (VAIC) does not indicate predictive value regarding performance (p > 0.05) and does not explain the variations in the Tobin coefficient.

3.3. Modern Performance Model

Taking into account the results of the model presented above, we continued our analysis by incorporating into the modern model the four dimensions identified as predictors of market performance, as shown in

Table 12. Result following selection for dependent values CU, CS, CR, ESG, according to the Summary model (modern model).

Model	R	R-Square	Adjusted R-Square	Std. Error of the Estimate
1	0.031 a	0.001	−0.009	4.207417204080182

^a Predictors: (Constant), ESG, CS, CR, CU.

.

Although the modern model illustrated in Table 12 does not demonstrate strong statistical significance, the integration of non-financial dimensions and sustainability factors provides a much more comprehensive conceptual perspective, and the value of intellectual capital is easier to assess compared to the traditional model.

Although the model in

Table 13. Result following selection for dependent values CU, CS, CR, ESG, according to the ANOVA model (modern model).

Model		Sum of Squares	df	Mean Square	F	Sig.
1	Regression	7.141	4	1.785	0.101	0.982 a
	Residual	7452.693	421	17.702
	Total	7459.834	425

^a Dependent variable: TQ. Predictors: (Constant), ESG, CS, CR, CU.

does not illustrate a strongly significant relationship with the market performance of the companies included in the sample, it covers a broader explanatory framework than the VAIC model.

To better illustrate the relationship between the identified dimensions and market performance, we present the estimated regression coefficients while also evaluating the individual contribution of each predictor, as shown in

Table 14. Estimation of the linear regression equation, according to the modern model.

Model		Unstandardized Coefficients		Standardized Coefficients	t	Sig.
		B	Std. Error	Beta
1	(Constant)	1.109	0.214		5.182	0.000
	CU	−4.270 × 10−7	0.000	−0.199	−0.401	0.689
	CS	1.161 × 10−5	0.000	0.013	0.257	0.798
	CR	0.000	0.008	−0.003	−0.054	0.957
	ESG	2.332 × 10−6	0.000	0.178	0.360	0.719

Dependent Variable: TQ.

.

Thus, we note that the inclusion of the four dimensions provides practitioners with a broader conceptual foundation and the opportunity to understand their contribution to the growth of intellectual capital value.

3.4. Exploratory MLP Analysis

The results show that high-quality human capital will have a positive effect on innovation, and companies will adapt much more easily to market trends (

Table 15. Independent variable: importance and influence of predictors on TQ.

	Importance	Normalized Importance
CU	0.373	100.0%
CS	0.121	32.6%
CR	0.343	92.1%
ESG	0.163	43.7%

).

This will implicitly increase the share price of companies listed on the BVB, in proportion to a higher market value. Our exploratory MLP analysis suggests that an increased value of HC may be associated with an increase in TQ, followed by the influence of relational capital with 92.1%, although these results should be interpreted with caution, given the exploratory nature of the analysis. This percentage confirms that employee relations are almost as important as human capital, especially for investors. The environmental, social, and governance factors have a positive impact on market value. Although IC or TQ values do not appear in the annual balance sheet, they are a defining element for value creation. To test the validity of the hypothesis that ESG has a significant influence on both IC and TQ, we applied an exploratory robustness analysis using the Multilayer Perceptron (MLP) model to illustrate the degree of importance of the four predictors and to capture the significance of sustainability factors. In this study, the model was applied to highlight the nonlinear relationships that emerge between company performance and the explanatory variables. This makes it much easier to highlight the indirect contribution these variables make to achieving high performance.

To assess the quality of the exploratory robustness analysis (MLP) and facilitate the comparison between the traditional and modern models,

Table 16. Exploratory robustness analysis (MLP).

Training	Sum of Squares Error	143.763
	Relative Error	0.998
	Stopping Rule Used	Relative change in training error criterion (0.0001) achieved
	Training Time	0:00:00.01
Testing	Sum of Squares Error	16.223
	Relative Error	0.994

Dependent Variable: TQ.

presents the performance indicators of the MLP for both the testing and training phases.

The results captured by the MLP also indicate high relative errors of 0.998 in the training phase and 0.994 in the testing phase. In our case, a value close to 1 suggests that the exploratory robustness analysis (MLP) does not adequately explain market performance measured through Tobin’s Q as the dependent variable. We acknowledge that a relative error of 0.998 reflects the difficulty of the MLP model in capturing market performance in this specific context. For this reason, we treat these results as exploratory, intended to illustrate the relative importance of predictors rather than to confirm the predictive accuracy of the model. Thus, we argue that there is a complex relationship between the dimensions of intellectual capital (IC) and the market value of the companies in our sample, which is influenced by external factors. One reason why market performance, through the dimensions CU, CS, CR, and ESG, is difficult to explain may be that the market of companies listed on the Romanian Stock Exchange does not sufficiently value intellectual capital. These results may also be considered a limitation of the proposed model, which could be addressed in future research by selecting a larger sample of companies.

4. Discussion

For a comparison between the tested models,

Table 17. Comparative analysis of the models.

Criterion	Traditional Model (VAIC)	Modern Model (CU, CS, CR, ESG)	Neural Network Model (MLP)	Interpretation
R2	0.000	0.001	—	Modern model has slightly higher explanatory capacity.
Adjusted R2	−0.002	-0.009	—	Identifies relevant dimensions.
ANOVA significance (p value)	0.919	0.982	—	Neither regression model is statistically significant.
Relative error (MLP)	—	—	0.994	The neural model confirms the difficulty of predicting TQ.
Relevant predictors	None	ESG (+), CS (+)	CU (100%), CR (92.1%)	Both models identify relevant dimensions
IC dimensions	Aggregated (VAIC)	CU, CS, CR, ESG	CU, CS, CR, ESG	The modern and neural models are multidimensional.
Includes ESG factors	No	✓	✓	Major advantage of modern models.
PCA composite score	No	✓	No	Enables latent measurement of IC.
Neural network analysis (MLP)	No	No	(exploratory)	Supports the robustness of the modern approach.
Captures intangible value	Limited	More comprehensive	More comprehensive	Conceptual advantage of modern models.

provides a synthesis of the main criteria analyzed in our study.

The results indicate that the conceptual framework is more comprehensive within the modern model. Although the statistical explanatory power of the modern model is not strong in terms of significance, it captures dimensions of intellectual capital that are not included in the traditional model. It incorporates ESG factors and allows for the latent measurement of intellectual capital through the application of the PCA composite score. Although our results indicate that the market of companies listed on the Bucharest Stock Exchange does not fully incorporate the value of intellectual capital into company prices, this does not mean that the model is not functional. Based on the information obtained, for the Romanian capital market, the recognition and transparency of the value of intangible assets remain limited.

The literature includes studies describing the relationship between human capital and intellectual capital value. Thus, a positive relationship exists between human capital efficiency and firm performance (Tran et al., 2020), and the use of an extended VAIC model can demonstrate that the human capital dimension exerts the greatest influence on intellectual capital (Evangelista et al., 2025).

When focusing specifically on the Romanian capital market, a study analyzing a sample of 64 companies listed on the BVB for the period 2016–2021 found a positive and statistically significant relationship between intellectual capital and corporate governance (Achim et al., 2023). Our study extended the scope of research by introducing ESG factors as a structural dimension of IC, as well as by using principal component analysis (PCA) to develop a composite index—a method not previously applied to BVB data. In the context of the literature reviewed by the authors, this is the first empirical study to test an extended intellectual capital model integrating ESG as a structural dimension on listed companies in Romania, covering a six-year panel dataset of 426 observations across nine sectors of activity.

Regarding the limited explanatory power of the traditional VAIC model in relation to market performance measured by Tobin’s Q, our findings confirm trends identified in previous studies. We thus conclude that VAIC does not have significant connections with the market value of the organization (Bataineh et al., 2022), and that the relationship between IC and market-based performance shows a low level of significance in emerging markets (Narula et al., 2024).

As for ESG factors, we find a positive, albeit limited, influence on the market value of the 75 companies analyzed. This is consistent with findings in the literature, where the influence of ESG on market performance, in emerging economies, is conditioned by contextual elements such as information transparency, investor awareness, and the regulatory framework (Handoyo & Anas, 2024).

Using the exploratory robustness analysis (MLP), we illustrated a classification of predictors according to their relative importance as follows: human capital 100%, relational capital 92.0%, ESG 43.7%, and structural capital 32.6%. The exploratory MLP analysis suggests that human capital appears to be the strongest predictor for the companies analyzed, a finding that should be interpreted with caution, given the limited statistical significance of the regression models.

The main objective of this study was to analyze intellectual capital by including non-financial dimensions and sustainability factors. Thus, the study examined which of the two approaches—traditional or modern—provides a more comprehensive explanatory framework. The analysis of the 75 companies listed on the Bucharest Stock Exchange included in the sample provided the opportunity to formulate some significant findings.

The quantitative results establish that the traditional model is limited in the context of IC assessment because it does not have the capacity to comprehensively cover the relationship between intellectual capital and market performance presented by Tobin’s Q (R² = 0.000, p = 0.919). Thus, the results offer support for hypothesis H1, established at the beginning of the study.

Although it has low explanatory power (R² = 0.001, p = 0.982), the modern model with the four integrated dimensions appears to provide a broader conceptual framework, which may represent an advantage in the process of understanding intellectual capital (IC). Hypothesis 2 receives partial conceptual support from the positive values associated with the ESG and structural capital (CS) dimensions, which contribute positively to the value of IC. Hypothesis 0 receives partial conceptual support, as the modern model provides a more realistic representation of IC in the current context of economic change. The modern model with the four integrated dimensions appears to provide a broader conceptual framework (Dsouza et al., 2025).

The relative importance of the predictors, as suggested by the exploratory MLP analysis, is as follows: human capital—100, relational capital—92.1, ESG factors—43.7, and structural capital—32.6. Market performance is affected by factors beyond the scope of the IC dimensions analyzed, which explains the high relative error recorded (0.994). The period covered by the study for the 75 companies is 2018–2023, characterized by multiple economic imbalances, such as the COVID-19 pandemic, which may represent another limitation of the study. Financial performance indicators may be influenced by political, economic, or social instabilities, making the presentation and interpretation of results more challenging. A future analysis of the same variables over a longer period and across a larger sample of companies could provide a clearer picture and stronger statistical significance. In this way, changes driven by the increasing integration of ESG factors and intellectual capital, alongside digital transformation, could be captured and explained more accurately.

5. Conclusions

This study focused on the analysis of intellectual capital by integrating non-financial dimensions and ESG factors into an extended evaluation model. The empirical basis of the analysis consists of a sample of 75 companies listed on the Bucharest Stock Exchange for the period 2018–2023. The main objective was to determine whether an integrative, modern model comprising four dimensions—human capital (CU), structural capital (CS), relational capital (CR), and ESG factors—offers a more complex explanatory perspective compared to the traditional VAIC model.

The traditional VAIC model has limited explanatory power in relation to market performance measured by Tobin’s Q. This was confirmed by the quantitative results (R² = 0.000; p = 0.919), which support hypothesis H1. The modern model presents a broader conceptual perspective, although it has low statistical significance (R² = 0.001; p = 0.982). It captures dimensions of intellectual capital that are not reflected in traditional models. These results represent arguments for the validation of hypothesis H2. Regarding hypothesis H0, it receives partial validation. The modern model does not bring major significance to explanatory power in statistical terms, but highlights a superior level of conceptual comprehension of intellectual capital in today’s economy.

In terms of predictor importance, they are as follows: human capital (100%), relational capital (92.1%), ESG factors (43.7%), and structural capital (32.6%). For the 75 companies analyzed, the exploratory MLP analysis suggests that the human dimension appears to play the most important role in the process of creating intangible value, although this finding should be interpreted with caution, given the exploratory nature of the MLP results. Another conclusion we draw from the results obtained refers to the fact that the Romanian capital market does not yet sufficiently incorporate the value of intellectual capital into company prices.

5.1. Limitations

We acknowledge that the study has several limitations. First, due to the limited availability of sustainability data, ESG indicators are constructed using proxy variables derived from publicly available financial statements. This situation generates a partial overlap between the components included in the study: the general solvency ratio appears both in the structural capital (CS) calculation and in the governance dimension (G) of the ESG score, while net profit per employee is used both in human capital (CU) and in the social dimension (S) of the ESG score. This overlap explains the high VIF values observed in the combined model (103.584 for CU and 103.499 for ESG) and represents an inherent limitation of working with data from emerging markets, where dedicated ESG databases are not systematically available.

Another limitation refers to the study period, 2018–2023, which is characterized by significant economic disruptions, including the COVID-19 pandemic. The COVID-19 pandemic brought significant instability to both financial performance and market values, and separating these effects from the real contribution of intellectual capital remains a challenge. These external shocks may have influenced financial performance indicators in ways that are difficult to isolate from the structural effects of intellectual capital, potentially affecting the stability and interpretability of the results. Data collection for the years 2024–2025 is currently underway, and future research will compare the results of the two periods.

Third, the sample is limited to 75 companies listed on the Bucharest Stock Exchange, which, although representative of the Romanian capital market, may not capture the full diversity of intellectual capital practices across different industries and company sizes. We are also aware that the limited number of companies listed on the Romanian capital market may make it harder to extend our findings beyond this specific context.

Fourth, the use of Tobin’s Q as the primary measure of market performance has known limitations, including sensitivity to share price fluctuations and limited comparability across industries (Dybvig & Warachka, 2015; Nawrocki, 2022).

5.2. Future Research Directions

Future research should address the limitations identified in this study. The use of standardized ESG data from dedicated databases—such as Bloomberg ESG, Refinitiv, or MSCI—would eliminate the overlap between components. This would produce more robust statistical results. Extending the analysis to a larger sample of companies, including unlisted companies or firms from other Central and Eastern European markets, would allow for broader generalizations and cross-market comparisons. The results of this research allow us to affirm that digital transformation and artificial intelligence represent emerging dimensions of intellectual capital that cannot be ignored. Current models, including the one proposed by us, do not fully capture this dimension. Future academic research can include these dimensions in extended models of intellectual capital evaluation.