CEOs' early famine experience, managerial discretion and corporate social responsibility

The study investigates whether and how CEOs' early-life famine experiences shape corporate social responsibility (CSR) performance. Grounded in imprinting theory and upper echelons theory, the paper posits that formative adverse experiences can leave enduring cognitive and value imprints that influence later strategic choices, including CSR. The context emphasizes rising societal expectations for CSR amid crises such as COVID-19 and recognizes that CEO characteristics significantly affect strategic outcomes. The research question centers on the direction and boundary conditions of the effect of CEOs' early famine experience on CSR within Chinese manufacturing listed firms (2010–2019), addressing mixed prior evidence and expanding understanding of antecedents to CSR.

Prior work links CSR to ownership structure, slack resources, stakeholder pressures, firm size/industry, and environmental dynamics. Upper echelons theory suggests executives' traits and life experiences affect strategic choices, with studies highlighting effects of military service, tenure, overseas experience, entrepreneurial orientation, and CEO duality on CSR. Imprinting theory argues that experiences during sensitive periods persistently shape behavior. Evidence on early-life hardship and CSR is mixed: some find negative links between famine experience and CSR, while others find positive relationships. Research on macro shocks (e.g., Great Depression) shows long-term effects on managerial conservatism and ethical behavior, suggesting early adversity may enhance precaution and prosocial orientation relevant to CSR.

Sample and data: Listed Chinese manufacturing firms from 2010–2019. Exclusions: S, ST, ST, SST, PT, indeterminate nature firms, and firms with serious missing data. Final dataset: 396 unbalanced panels with 2,202 firm-year observations. Continuous variables winsorized at 1% and 99%. Data sources: CSMAR, Runling Global, Hexun.com, Compilation of Historical Statistical Data of Provinces (1949–1989), and others. Dependent variable (CSR): CSR performance measured by RKS CSR report ratings (based on GRI framework adapted to China). Higher scores indicate better CSR. Independent variable (CFE): CEO early famine experience. CEOs born such that their childhood (0–14) overlapped with 1959–1961 Great Famine are coded 1; otherwise 0. Baseline coding marks CFE=1 if CEO birth year 1947–1961; robustness uses alternative time windows/cohorts. Moderators:

• Market abundance (MA): environmental discretion proxy; measured using the industry’s average sales growth/variation over the past five years (coefficient of variation). Higher values denote richer market munificence.
• Slack resources (OS): organizational discretion proxy; measured by equity-to-debt ratio (higher indicates more slack).
• CEO concurrently (CI): CEO discretion proxy; dummy=1 if CEO also chairs the board (duality), 0 otherwise. Controls: Firm size (log assets), firm age (log years), leverage (assets-to-liabilities ratio), market share (firm main business revenue share of industry), ROE, growth (revenue growth), equity nature (SOE=1), CEO gender (male=1), CEO age (log), CEO education (ordinal: 1–5 from below junior college to doctorate), CEO turnover (1 if departed), plus year and industry fixed effects. Models: Panel regressions with Driscoll–Kraay standard errors. (1) CSR = β0 + β1 ΣControls + ε (2) CSR = β0 + β1 CFE + β2 ΣControls + ε (3) CSR = β0 + β1 CFE + β2 (CFE×MA) + β3 MA + β4 ΣControls + ε (4) CSR = β0 + β1 CFE + β2 (CFE×OS) + β3 OS + β4 ΣControls + ε (5) CSR = β0 + β1 CFE + β2 (CFE×CI) + β3 CI + β4 ΣControls + ε Endogeneity and robustness: System GMM for dynamic panels (lagged CSR, Hansen J, AR(1)/AR(2) diagnostics); IV approach using famine-affected population (FAP) as instrument; alternative CSR measure (Hexun CSR ratings); alternative CFE cohort definitions (infancy, early childhood, childhood/adolescence, adult); heterogeneity analyses by ownership (SOE vs non-SOE), CEO turnover, CEO gender, and CEO education level.

• Main effect: CEOs with early famine experience are associated with higher CSR performance. Regression coefficient: CFE Beta ≈ 0.682 (p<0.05).
• Environmental discretion (market abundance): Positive moderation. Interaction CFE×MA Beta ≈ 13.009 (p<0.01); effect remains in full model.
• Organizational discretion (slack resources): Positive moderation. Interaction CFE×OS Beta ≈ 1.497 (p<0.05); robust in full model.
• CEO discretion (duality): Negative moderation. Interaction CFE×CI Beta ≈ -0.850 (p<0.05); in alternative robustness some specifications render this interaction not significant.
• Descriptive statistics: Mean CSR 37.94 (SD 10.94). Mean CFE 0.215 (SD 0.411).
• Endogeneity checks: System GMM results pass Hansen J and AR tests; main and moderation effects broadly consistent (e.g., CFE×MA positive and significant; CFE×OS positive in some GMM specs; CI effects negative).
• IV (FAP) robustness: Instrument significant; results consistent with main findings except the CEO discretion cross-term not significant in some IV models.
• Alternative CSR measure (Hexun): Results consistent for main effect and positive moderations; CEO discretion interaction often not significant.
• Alternative CFE cohorts: Results broadly consistent; organizational discretion interaction sometimes insignificant depending on specification.
• Heterogeneity:
  • Ownership: Stronger positive effect in SOEs (CFE Beta ≈ 0.587, p<0.05) than in non-SOEs.
  • CEO turnover: Reappointed CEOs show significant positive effect (CFE Beta ≈ 0.806, p<0.05); turnover group not significant.
  • Gender: Effect stronger and significant for male CEOs (CFE Beta ≈ 0.661, p<0.01); female group not significant.
  • Education: Stronger effect for CEOs without higher education (below bachelor) (CFE Beta ≈ 5.494, p<0.01); not significant for higher-educated group.

Findings support imprinting theory and upper echelons theory: early-life famine imprints foster precaution, empathy, and long-term orientation, translating into higher CSR engagement. Environmental and organizational discretion amplify the ability to enact CSR by supplying external opportunities and internal slack, buffering uncertainty and enabling resource-intensive CSR initiatives. In contrast, concentrated CEO power via duality may weaken governance checks and stakeholder orientation, dampening the positive imprint effect on CSR. Heterogeneity patterns align with institutional and personal contexts: SOEs' mission orientation and access to resources reinforce the effect; reappointed CEOs can more readily channel imprinted values into CSR given firm-specific knowledge; male CEOs and those without higher education display stronger imprint-behavior links, potentially due to cultural and motivational factors influencing risk, reputation, and prosocial actions.

The study shows that CEOs' early famine experiences are positively associated with corporate CSR performance in Chinese manufacturing firms. Environmental (market abundance) and organizational (slack resources) discretion strengthen this relationship, while CEO discretion via duality weakens it. The positive effect is more pronounced in SOEs, among reappointed CEOs, male CEOs, and CEOs without higher education. Contributions include extending imprinting and upper echelons perspectives to CSR antecedents, identifying managerial discretion as a key boundary condition, and providing context-specific evidence from China. Managerially, boards and policymakers should consider leadership life-history imprints and ensure enabling discretion (external munificence, internal slack) while avoiding excessive concentration of CEO power to foster CSR. Future research should validate these findings across broader samples and settings, explore longer time horizons, incorporate additional contextual moderators (e.g., environmental dynamics and institutional systems), and examine leadership styles as further mechanisms.

The sample focuses on Chinese listed manufacturing firms, limiting generalizability to non-listed or other sectors. The panel spans 2010–2019, a relatively short horizon for imprint effects. Despite extensive controls, omitted variables may remain. Some moderating effects lose significance under alternative measures/instruments. Additional boundary conditions such as environmental dynamism and institutional systems were not fully examined. Future work should broaden samples and periods, add controls and contextual moderators, and investigate leadership styles as mechanisms.