Does intellectual capital impact the financial performance of Indian public sector banks? An empirical analysis using GMM

The study investigates whether and how intellectual capital (IC) influences the financial performance (FP) of Indian public sector banks (IPSBs). Motivated by the increasing importance of intangible resources in knowledge-intensive service sectors like banking, and grounded in resource-based theory, the paper highlights IC—comprising human, structural, relational, and capital employed elements—as a strategic asset for competitive advantage and sustainability. India’s rapid economic growth, financial sector digitization, and policy reforms underscore the need to understand IC’s role in bank profitability and stability. Prior research offers mixed evidence on the IC–performance nexus and has largely relied on simpler econometric approaches. This study contributes by applying the MVAIC framework to capture IC comprehensively and using a dynamic panel GMM estimator to address endogeneity and the volatile nature of bank earnings, covering a long horizon (2010–2021) that includes the COVID-19 period. The central research question is the extent to which IC and its components affect accounting- and market-based performance metrics (ROA, ROE, ROCE, EPS, Tobin’s Q) of IPSBs.

The literature conceptualizes intellectual capital (IC) as knowledge-based intangible resources—knowledge, expertise, processes, relationships, and information—that create organizational value. IC is commonly decomposed into human capital (HC), structural capital (SC), relational capital (RC), and capital employed (CE). HC encompasses employees’ skills and competencies; SC includes organizational structures, processes, systems, and culture; RC represents value embedded in stakeholder relationships; CE captures the efficiency of financial and physical resources. Multiple models assess IC, including the VAIC, MVAIC, Balanced Scorecard, Skandia Navigator, and others. VAIC has been widely used due to data availability but omits RC and innovation; MVAIC extends VAIC by adding RC, providing broader coverage of IC components. Empirical evidence on IC’s effect on performance is mixed: many studies report positive associations between IC (and especially HC and SC) and performance across countries and sectors, while others find insignificant or varied effects by component. Studies focusing on banking often show IC’s importance for profitability and efficiency, yet methodological limitations (e.g., OLS, lack of dynamic controls, endogeneity concerns) remain. Recent works advocate dynamic panel estimators (e.g., GMM). In the Indian context, few studies explore banking IC using comprehensive models and robust estimators; this study addresses that gap and includes the COVID-19 period.

Data: The sample comprises 23 Indian public sector banks (before and after mergers), 2010–2021, forming an unbalanced panel. Financial data were sourced primarily from CMIE ProwessIQ; missing observations were supplemented from banks’ annual reports. Banks with irrecoverable missing data were excluded. IC measurement: The study employs the Modified Value-Added Intellectual Coefficient (MVAIC), aggregating efficiencies of HC, SC, RC, and CE: MVAIC = HCE + SCE + RCE + CEE. Component definitions: - HCE = VA / HC, where HC is total personnel expense. - SCE = SC / VA = (VA − HC) / VA. - CEE = VA / CE, where CE denotes financial and physical capital employed. - RCE = RC / VA, where RC is proxied by marketing, sales, and advertising expenses. Value added (VA) = Operating profit + Employee cost + Depreciation + Amortization. Dependent variables (financial performance): - ROA = Net income / Total assets - ROE = Profit to equity shareholders / Shareholders’ funds - ROCE = EBIT / Capital employed - EPS = Net income / Shares outstanding - Tobin’s Q (TQ) = (Market value of equity + Book value of short-term liabilities) / Book value of total assets Controls: Leverage (total outside liabilities / total assets) and Size (log of total assets). Pre-estimation diagnostics: Descriptive statistics and Pearson correlations were computed. Panel unit root tests (LLC; Fisher-ADF; Fisher-PP) indicate non-stationarity at level and stationarity at first difference; thus all series were differenced once (Δ) prior to estimation. Estimation strategy: Two-step Arellano–Bond GMM for dynamic panels was used to address endogeneity, firm-specific effects, and potential heteroscedasticity. Models include the lagged dependent variable as a regressor, IC components (CEE, HCE, SCE, RCE) in Models 1–5 and MVAIC in Models 6–10, along with controls (Leverage, Size). Instrument validity and specification were assessed via: - Arellano–Bond tests for AR(1) and AR(2) in differences (absence of second-order serial correlation expected), and - Sargan test (J-statistic) for overidentifying restrictions (p>0.05 indicates valid instruments). Model set: - Models 1–5: ΔROA, ΔROE, ΔEPS, ΔTQ, ΔROCE each regressed on their lag, CEE, HCE, SCE, RCE, Leverage, Size. - Models 6–10: ΔROA, ΔROE, ΔEPS, ΔTQ, ΔROCE each regressed on their lag, MVAIC, Leverage, Size.

Descriptive patterns: - Among IC components, HCE shows the highest mean efficiency (mean HCE = 1.888), and average MVAIC = 3.672. Non-physical IC components (HCE, RCE, SCE) jointly average 3.547 versus CEE = 0.123, indicating greater value creation from intangibles than from physical/financial capital. - EPS exhibits a high mean (18.919), ROCE averages 1.391, and TQ < 1 on average, suggesting market value below book value during the period. GMM results with IC components (Models 1–5): - Human Capital Efficiency (HCE): Positively and significantly associated with ROA, ROE, EPS, and ROCE; negatively associated with TQ. - Capital Employed Efficiency (CEE): Negatively and significantly related to ROA, ROE, EPS, and ROCE; positively associated with TQ, implying CEE supports market valuation though not accounting profitability. - Structural Capital Efficiency (SCE): Negatively associated with ROA (significant) and ROE/ROCE (generally negative, with varying significance); weak/insignificant for EPS and TQ. - Relational Capital Efficiency (RCE): Negatively associated with EPS (significant); positively associated with ROCE (significant), and generally insignificant for ROA, ROE, and TQ. - Controls: Leverage exhibits a negative and significant effect across performance metrics; Size effects are generally weak/insignificant. - Dynamics and diagnostics: Lagged dependent variables are significant. AR(1) sometimes present but AR(2) absent, indicating no problematic serial correlation; Sargan tests (p>0.05) support instrument validity. GMM results with MVAIC (Models 6–10): - MVAIC shows a positive and significant influence on ROA and ROE; effects on EPS, ROCE, and TQ are statistically weaker, with TQ directionally negative. - Leverage remains negatively and significantly related to all five FP indicators. - Diagnostic tests confirm valid instruments and no second-order serial correlation. Overall: IC matters for IPSBs’ performance. HCE emerges as the principal driver of accounting-based profitability, while CEE relates positively to market-based performance (TQ) but negatively to accounting returns. Higher leverage consistently erodes performance.

The results directly address the research question by demonstrating that intellectual capital significantly influences IPSBs’ performance, with heterogeneous effects by component. Human capital efficiency is the most potent positive contributor to accounting-based FP (ROA, ROE, EPS, ROCE), underscoring the role of employee skills, expertise, and capacity for innovation in a knowledge-intensive banking context. Conversely, structural capital and capital employed efficiencies show negative relations with several accounting performance measures, suggesting inefficiencies in the deployment of physical/financial capital and internal processes. However, capital employed efficiency aligns positively with market valuation (TQ), indicating that the market rewards efficient use of invested capital even when accounting returns lag. The combined IC measure (MVAIC) reinforces that overall IC accumulation and deployment enhances core profitability (ROA, ROE). The consistently negative effect of leverage indicates that greater debt burdens reduce bank value and profitability, likely through higher interest costs and elevated financial risk. Robustness checks via GMM diagnostics support the reliability of the estimates, with valid instruments and absence of problematic second-order serial correlation. Collectively, the findings affirm IC’s centrality in driving sustainable bank performance and guide prioritization among IC components—particularly investment in human capital—within IPSBs.

This study contributes to the IC–performance literature by applying a comprehensive MVAIC framework and dynamic GMM estimation to Indian public sector banks over 2010–2021, including the COVID-19 period. It shows that IC, especially human capital efficiency, materially improves accounting-based performance, while capital employed efficiency is positively linked to market-based valuation but detrimental to several accounting measures. Aggregate IC (MVAIC) significantly enhances ROA and ROE. Policy and managerial implications include prioritizing investments in human capital (training, development, and retention), reassessing the deployment of physical/financial capital and structural processes to reduce inefficiencies, and managing leverage prudently to mitigate adverse effects on performance. Future research directions include expanding the sample to private, cooperative, rural, payment, and small finance banks; examining non-banking industries; incorporating additional performance metrics; and comparing alternative IC measurement frameworks beyond MVAIC (e.g., national IC index, intangible assets monitor, balanced scorecard, VAIC).

- Scope limited to Indian public sector banks; results may not generalize to private or other financial institutions and non-banking sectors. - Data constraints led to an unbalanced panel; some banks with irrecoverable missing data were excluded, potentially introducing selection bias. - IC measurement relies on the MVAIC approach and proxy definitions (e.g., RC via marketing/sales/advertising expenses), which may not capture all dimensions of intellectual capital (e.g., innovation capital). - Variables are differenced to address non-stationarity, potentially affecting interpretability of levels relationships. - While GMM addresses endogeneity and dynamics, results depend on instrument validity and standard assumptions (e.g., no second-order serial correlation); finite-sample instrument proliferation risk remains. - The period includes structural changes (e.g., mergers, COVID-19 shock) that may influence outcomes beyond modeled controls.