Value relevance of multifaceted corporate social performance: how do country-specific factors matter?

There is growing interest in socially responsible investment and how corporate social performance (CSP) relates to shareholder value. Stakeholder theory suggests that integrating diverse stakeholder demands can create added value for shareholders, while shareholder theory cautions that social activities may impose costs that erode value. Prior findings on CSP’s valuation effect are mixed, potentially due to aggregation bias, model misspecification, and measurement error in CSP. This study asks whether different types of corporate social activities aimed at distinct stakeholder groups create shareholder value and how country-specific factors moderate this relationship. Contributions include: (1) examining disaggregated CSP dimensions (community, human rights, product responsibility, workforce) to avoid aggregation bias; (2) assessing moderating roles of economic development, political system (corruption), and culture (humane orientation); (3) applying importance-performance analysis (IPA) to prioritize CSP dimensions under limited resources; and (4) addressing CSP measurement error by employing structural equation modeling (SEM) rather than relying solely on regression. Using Asian-Pacific emerging market data (2010–2020), the study finds that aggregate CSP lacks relevance in regressions, but disaggregated dimensions—especially workforce—are value-relevant; SEM reveals synergistic positive effects of all CSP dimensions on shareholder value. The valuation effect is stronger in countries with higher income per capita and lower corruption. IPA recommends prioritizing human rights, then community, product responsibility, and workforce. The paper outlines literature, hypotheses (H1–H4), analytical framework, dataset, empirical results, and implications for policy and management.

The literature emphasizes measuring CSP with multidimensional indicators to address aggregation bias. Studies examining distinct CSP dimensions (e.g., community, workforce relations, diversity, environment, product responsibility, governance, human rights) show heterogeneous effects on firm risk and financial distress (Salama et al., 2011; Bouslah et al., 2013; Boubaker et al., 2020). Shareholder value is commonly measured by financial metrics such as earnings yield, ROE, P/E, P/B, dividends, and free cash flow; earnings yield is a robust valuation metric, avoiding pitfalls of P/E near-zero earnings (Eun and Lee, 2010; Musumeci and Peterson, 2011). Theoretical perspectives diverge: stakeholder theory argues CSP can enhance reputation and attract stakeholder resources, improving profitability and value; shareholder theory warns CSP may divert resources and harm performance. Empirical evidence on CSP-value links is mixed, with potential explanations including measurement unreliability, endogeneity, omitted variables, and model misspecification. Endogeneity-adjusted studies sometimes find diminished or negative links (Garcia-Castro et al., 2010; Lahouel et al., 2019). CSP can affect value through earnings growth and risk channels; visible CSP can broaden investor base. Research has focused more on developed markets, with fewer studies in emerging markets where institutional contexts differ, suggesting findings may not generalize due to regulatory, cultural, and economic differences. This motivates a nuanced, multidimensional analysis of CSP with attention to country-level moderators.

Analytical framework combines panel regression and structural equation modeling (SEM) to assess CSP value relevance while accounting for measurement error and country moderators.

• Regression model (fixed effects): ENY_it = β1 CSP_it + γ CTL_it + F_i + Y_t + ε_it, where ENY is earnings yield (stock value proxy), CSP is either aggregate social pillar score (SOC) or category scores (SOC1 community, SOC2 human rights, SOC3 product responsibility, SOC4 workforce). Controls: dividend payout ratio (DPO), earnings growth (GRO), risk-free rate (RFR, 3-month T-bill), equity risk premium (ERP, proxied by stock beta), market capitalization (MCP), debt-to-asset ratio (DAR). Expectations: ENY positively related to RFR, ERP, DAR; negatively related to CSP, DPO, GRO, MCP. Firm and year fixed effects control for time-invariant firm heterogeneity and common time shocks.
• SEM for CSP as latent construct: Treat CSP as latent variable measured by SOC1–SOC4 to separate measurement error from structural error. Construct equation: ENY_it = β1 CSP_it + γ CTL_it + F_i + Y_t + ε_it. Measurement equations: SOCk_it = θ_k + ω_k CSP_it + μ_kit for k=1..4. Factor loadings expected positive. Identification via fixing one loading to unity in alternative models (SOC1–SOC4 each used in turn). Goodness-of-fit assessed via CFI and RMSEA.
• Moderating roles of country factors (SEM extension): Include moderator (MOD) and interaction CSP×MOD in construct equation: ENY_it = β1 CSP_it + β2 MOD_it + β3 CSP_it×MOD_it + γ CTL_it + F_i + Y_t + ε_it. Moderators: economic development (natural log of income per capita, Ln(IPC)), political system (corruption perception index, CPI; higher implies less corruption), and culture (humane orientation, HUM; binary: 1 for Southern Asia: India, Indonesia, Malaysia, Philippines, Thailand; 0 for Confucian Asia: China, Hong Kong, Taiwan). Expect β3<0 (since higher value corresponds to lower earnings yield). Due to CPI methodology changes post-2012, moderation analyses use 2012–2020 subsample.
• Importance-Performance Analysis (IPA): Importance proxied by standardized factor loadings from SEM (SOC1–SOC4). Performance proxied by medians of each category score. Compute performance-to-importance ratio to prioritize improvements: lower ratio indicates higher priority. Plot importance vs performance to guide resource allocation.
• Data: CSP measures from Refinitiv Eikon ESG dataset (social pillar: workforce 30 metrics; human rights 8; community 14; product responsibility 10; pillar weights 0.28, 0.17, 0.13, 0.43; scores 0–100 normalized with median ~50). Sample: 668 firms, 5,626 firm-year observations across 8 Asia-Pacific emerging markets (China, Hong Kong, India, Indonesia, Malaysia, Philippines, Taiwan, Thailand), 2010–2020. Financial variables from Refinitiv Eikon. Country variables: income per capita (Refinitiv), CPI (Transparency International), HUM from GLOBE clusters. Stationarity verified with panel ADF tests. Regression uses robust t-stats; SEM uses ML with robust z-stats; firm and year fixed effects included throughout.

• Descriptive CSP levels: Except Thailand and India, median social pillar scores are below 50 across markets, indicating underperformance vs. the Refinitiv ESG universe; human rights is consistently the weakest category.
• Panel regression (fixed effects):
  • Aggregate CSP (SOC) shows expected negative sign but is statistically insignificant, suggesting aggregation bias can mask effects.
  • Examining categories separately (Models 3–6): all coefficients are negative; community and workforce are significant (community at 10%, workforce at 5%), human rights and product responsibility are not significant.
  • Including all categories jointly (Model 7): only workforce remains significantly negative (e.g., workforce coefficient around -0.041 with significance), indicating heterogeneous market valuation of CSP dimensions.
• SEM treating CSP as latent variable (Models 8–11):
  • Measurement: All factor loadings positive and highly significant; standardized loadings: workforce 0.962, community 0.900, product responsibility 0.896, human rights 0.782.
  • Construct: CSP latent variable significantly negatively affects earnings yield in all models (coefficients approx. -0.376 to -0.194; z ≈ -3.2 to -3.4), implying a positive effect on stock value when accounting for measurement error and synergy among dimensions.
  • Goodness-of-fit: CFI ≈ 0.968; RMSEA ≈ 0.042; adjusted R2 ≈ 0.558, exceeding OLS models (≈0.462–0.470).
• Moderation by country factors (SEM, 2012–2020; Models 12–15):
  • Economic development (Ln(IPC)): CSP×Ln(IPC) negative and significant (e.g., -0.021**, -0.019**), indicating stronger CSP–value link in higher-income countries (H4A supported).
  • Political system (CPI, higher=less corruption): CSP×CPI negative and marginally significant (≈ -0.010, p<0.10; -0.009*, p<0.10), indicating stronger CSP–value link where corruption is lower (H4B supported).
  • Culture (HUM): CSP×HUM negative but not statistically significant; no moderating effect detected (H4C not supported).
• Importance-Performance Analysis (IPA):
  • Performance medians: human rights 10.880; community 32.425; product responsibility 43.420; workforce 54.385.
  • Importance (standardized loadings): workforce 0.962; community 0.900; product responsibility 0.896; human rights 0.782.
  • Performance-to-importance ratios: human rights 13.913 (highest priority), community 36.028 (2nd), product responsibility 48.460 (3rd), workforce 56.533 (4th). Overall: Aggregate CSP appears value-irrelevant in OLS due to aggregation and measurement issues; disaggregated and SEM evidence shows CSP—especially narrowly focused dimensions—enhances shareholder value, with effects stronger in higher-income and lower-corruption contexts.

Findings demonstrate that CSP’s value relevance is multidimensional and context-dependent. The inconclusive or insignificant effects of aggregate CSP in regression are attributable to aggregation bias and measurement error; SEM reveals a robust, synergistic positive impact of CSP on stock value, consistent with stakeholder theory. Narrowly focused CSP (community, product responsibility, workforce) produces more tangible benefits to primary stakeholders—enhancing productivity, reputation, customer trust, and reducing litigation and operational risks—thereby improving expected cash flows and lowering cost of capital. Broadly focused CSP (human rights) shows weaker direct valuation effects in regressions, likely reflecting less immediate, localized stakeholder benefits; however, SEM indicates it still contributes positively through synergy with other dimensions. Country context matters: in higher-income and lower-corruption environments, CSP’s benefits are more visible, enforceable, and valued by markets, strengthening the CSP–value link. No moderating effect is found for humane orientation, suggesting cultural clusters (as measured) may not capture variation relevant to financial market valuation of CSP. Managerially, firms should pursue a balanced CSP portfolio leveraging synergies, while prioritizing improvements where performance lags relative to importance (human rights first), and tailoring expectations to institutional environments where CSP is more highly rewarded by markets.

The study advances understanding of CSP’s value relevance by: (1) demonstrating that aggregate CSP can obscure heterogeneous valuation effects across dimensions; (2) showing via SEM that all CSP dimensions jointly and significantly enhance stock value, with stronger effects for narrowly focused dimensions; and (3) documenting that economic development and lower corruption strengthen CSP’s valuation impact, while humane orientation does not. Practically, managers should not rely solely on aggregate CSP scores; instead, manage a portfolio of CSP initiatives and prioritize improvements using IPA: human rights, community, product responsibility, and workforce, in that order. Policymakers in emerging markets can enhance CSP’s value impact by strengthening institutions and anti-corruption efforts. Future research should extend the framework to include environmental and governance pillars, explore additional moderators (e.g., political uncertainty, degree of multinationality, regulatory stringency), and test generalizability beyond Asian-Pacific emerging markets and the 2010–2020 window.

• Scope limited to social pillar of ESG; environmental and governance pillars are excluded, potentially omitting important interactions across ESG dimensions.
• Cultural moderator measured via broad GLOBE clusters (HUM) may be too coarse to detect nuanced cultural effects.
• Moderation analysis constrained to 2012–2020 due to CPI methodology changes, reducing sample size and possibly statistical power.
• Sample limited to 8 Asia-Pacific emerging markets and 2010–2020; results may not generalize to developed markets or other regions/periods.
• SEM addresses measurement error in CSP indicators but cannot eliminate all endogeneity concerns (e.g., time-varying omitted variables beyond fixed effects).
• Unbalanced panel with occasional data gaps; reliance on proprietary Refinitiv Eikon data may limit replication by those without access.