The nexus of top executives' attributes, firm strategies, and outcomes: Large firms versus SMEs

The study investigates whether, which, when, and to what extent top executives influence firm strategic decision-making and performance. Building on upper echelons theory (UET), which posits that organizational outcomes reflect the values and cognitive bases of top executives, the paper addresses the limitations of demographic proxies (the “black box problem”) by incorporating psychological attributes. It emphasizes the contextual nature of managerial discretion, noting cultural and organizational factors, and highlights Japan as a context where CEO effects are often viewed as low. The research aims to compare large firms and SMEs in Japan, test whether strategic choices mediate the relationship between executives’ attributes and performance, and examine how these relationships vary by firm size. The central hypothesis is that top executives’ attributes have stronger impacts in SMEs than in large firms through strategic choices (financial leverage and R&D investment) and on firm outcomes.

Theoretical background centers on UET (Hambrick and Mason, 1984) and subsequent evidence that top managers affect strategy and performance. Prior work has used manager fixed effects and demographic proxies (age, education, experience) to infer executive impacts but faces imprecision due to lack of direct psychological assessment. Psychological constructs (e.g., overconfidence, narcissism, Big Five) have been linked to corporate risk and policies, though data collection is challenging, especially for SMEs. Managerial discretion varies across macro (national culture, institutions) and micro (firm size, ownership, strategy type) contexts. Cross-country studies suggest lower CEO discretion and performance effects in Japan. Within organizations, larger/older firms face inertia; SMEs’ executives (often owners) may wield greater discretion. Strategy type also matters; strategic risk choices (leverage, R&D) are common across firm sizes and amenable to executive influence. Prior SME studies show strategic change/flexibility mediate personality–performance links. This study addresses gaps by using psychological assessments for a broad sample including SMEs and by comparing large firms versus SMEs in Japan.

Data: A corporate credit survey from Teikoku Databank Ltd. assessed top executives (typically the firm president/representative director) via trained researchers’ interviews and head office confirmation. There are 970,206 executive observations (1985–2014) with 25 binary indicators covering personal traits (e.g., seriousness, steadiness, tolerance, kindness) and managerial skills (e.g., execution, extraversion, technology orientation). Exploratory factor analysis (oblimin rotation; parallel analysis) yielded 9 factors: Big Five (conscientiousness, extraversion, neuroticism, openness, agreeableness) and four attributes (uniqueness, execution skills, decisiveness, financial prudence). Factor loadings align with conceptual definitions (e.g., conscientiousness loads on serious/steady/responsibility; financial prudence loads on thrifty/prudent/attention to detail).

Financial and firm data: One-year post-survey financials form a pooled cross-section (2010–2014) of 85,175 firm-year observations after exclusions (financial sector; sectors with <10 firms; missing values). Performance is ROA (net income/total assets). Strategic mediators: financial leverage = (total liabilities − cash)/total assets; R&D intensity = R&D expenditure/revenue (missing R&D set to zero). Controls include firm size (log assets), lagged profitability (ROA), historical revenue growth (3-year average), firm age, tangibility (fixed assets/total assets), and capital intensity (log fixed assets/employees). Fixed effects: sector, year, location, listing status. Firm types (large, medium, small) follow Japan’s SME Basic Act; 93.58% of the sample are SMEs (57.10% medium, 36.48% small).

Model: Structural equation modeling (SEM) tests mediation and heterogeneity across firm types. Equations: (1) ROA regressed on factors and mediators; (2) leverage on factors; (3) R&D intensity on factors; with appropriate controls per equation. Estimation uses diagonally weighted least squares (DWLS) due to non-normality. Model fit indices (CFI, TLI, RMSEA, SRMR) are reported for full and subgroup samples. Multigroup analyses and chi-square difference tests assess path coefficient invariance across firm types to identify where effects differ (factors→ROA; factors→mediators; mediators→ROA).

• Model fit: Good fit overall (full sample: CFI=0.992, TLI=0.891, RMSEA=0.014, SRMR=0.004). Subsamples: Large (CFI=0.993, RMSEA=0.020, SRMR=0.011), Medium (CFI=0.983, RMSEA=0.021, SRMR=0.007), Small (CFI=0.995, RMSEA=0.010, SRMR=0.003).
• Mediators to performance: Both financial leverage and R&D intensity negatively relate to ROA across firm types (e.g., full sample: leverage b≈−0.068, p<0.01; R&D intensity b≈−0.247, p<0.01).
• Direct effects on ROA (full sample): Neuroticism (b=0.223, p=0.002) and agreeableness (b=0.270, p<0.001) positive; decisiveness negative (b=−0.186, p=0.01). By firm size: In small firms, neuroticism (b=0.553, p<0.01) and agreeableness (b=0.556, p<0.01) positive; decisiveness negative (b=−0.604, p<0.01). In large firms, decisiveness positive (b=0.486, p<0.05); in medium firms, no significant factor-to-ROA direct effects at p<0.05.
• Factors to financial leverage (full sample): Neuroticism (b=2.093, p<0.001), openness (b=1.713, p<0.001), agreeableness (b=3.090, p<0.001) increase leverage; conscientiousness (b=−1.567, p<0.001), decisiveness (b=−1.300, p<0.001), and financial prudence (b=−4.278, p<0.001) decrease leverage. Magnitudes are generally strongest in small firms (e.g., financial prudence b=−6.339 in small firms vs −2.790 in medium; not significant in large). Agreeableness has strongest positive effect on leverage (small firms b=4.596).
• Factors to R&D intensity (full sample): Openness strongly increases R&D (b=0.176, p<0.001), with larger effects in large (b=0.276, p<0.05) and small firms (b=0.280, p<0.01) than medium (b=0.117, p<0.001). Neuroticism (b=0.037, p<0.05) and agreeableness (b=0.040, p<0.01) also positive. Execution skills (b=−0.059, p<0.01), extraversion (b=−0.054, p<0.01), and decisiveness (b=−0.049, p<0.01) negatively relate to R&D intensity.
• Mediation: Strategic choices significantly mediate attribute–performance links. Indirect effects via financial leverage are generally stronger than via R&D. Conscientiousness, decisiveness, and financial prudence have positive indirect effects on ROA by lowering leverage; uniqueness, neuroticism, openness, and agreeableness have negative indirect effects by raising leverage. For small firms: a one SD increase in financial prudence raises ROA by ~0.212 pp via lower leverage (≈13.6% of mean ROA), while a one SD increase in openness lowers ROA by ~0.246 pp via higher leverage (≈15.9% of mean ROA).
• Firm-size heterogeneity: Executive attributes weakly predict strategies/outcomes in large firms; effects are stronger in SMEs, especially small firms, for both direct and indirect paths. Path invariance tests show factor→mediator paths differ significantly across firm sizes for most attributes; mediator→ROA paths do not differ significantly across sizes.
• Inconsistent mediation: In SMEs (notably small firms), opposing signs between direct and indirect effects attenuate total effects (e.g., neuroticism and agreeableness have positive direct but negative indirect effects), helping explain previously observed low CEO performance effects in Japan.

Findings support that top executives matter more in SMEs than in large firms due to greater managerial discretion. In large firms, performance is less sensitive to executive attributes; notable effects include a positive direct effect of decisiveness and a positive indirect effect of conscientiousness via lower leverage. In medium-sized firms, effects on performance operate mostly through strategy (leverage, R&D), with conservative attributes (conscientiousness, decisiveness, prudence) associated with lower leverage and risky attributes (neuroticism, openness, agreeableness) with higher leverage and R&D. In small firms, executives’ attributes exert both direct and indirect effects: neuroticism and agreeableness positively affect ROA directly, while several attributes influence leverage and R&D, producing substantial indirect effects. However, inconsistent mediation (opposing direct and indirect signs) often reduces total observable performance effects, reconciling prior evidence of low CEO effects in Japan. Differences in pathways arise primarily from attribute→strategy links rather than strategy→performance links. Organizational context (ownership/founder or family successor prevalence in SMEs vs internal promotions in large firms) likely underpins discretion differences; network structures (e.g., keiretsu) may further shape heterogeneity across firm sizes.

The study shows that top executives’ psychological attributes shape firm strategic risk choices and, through these, performance, with stronger effects in SMEs than in large firms in Japan. Conscientiousness, decisiveness, and financial prudence relate to conservative financial policies; neuroticism, openness, and agreeableness relate to riskier financial policies, and openness is a key driver of R&D intensity. Mediation via leverage is generally stronger than via R&D. The heterogeneity across firm sizes highlights nuanced managerial discretion: small-firm executives leave a stronger imprint on strategies and outcomes, though inconsistent mediation may mask total performance effects. Contributions include a multidimensional measure of executive attributes from a large-scale psychological dataset, comparative evidence across firm sizes, and illumination of mediation mechanisms in the Japanese context. Future research should expand to cross-national comparisons, incorporate richer contextual typologies, and use longitudinal designs to assess long-term and causal effects of executive attributes on strategy and performance.

• External validity: The sample comprises only Japanese firms; cultural and institutional specificity may limit generalizability.
• Context measurement: One-dimensional contextual factors may not fully capture the complexity of managerial discretion; multidimensional context typologies could improve precision.
• Research design: Cross-sectional financial data (2010–2014) linked to executive assessments preclude causal inference and omit long-term performance dynamics valued in Japan; longitudinal data are needed.
• Data constraints: R&D intensity uses zeros for missing values; psychological measures rely on researcher assessments from interviews, which, while systematic, may introduce measurement noise.