Are optimistic CEOs and pessimistic CFOs the best partners? Evidence from corporate cash holdings

The study investigates whether an optimistic CEO combined with a pessimistic CFO forms an effective executive pairing for corporate financial policy, focusing on cash holding decisions. CEOs lead strategy and often exhibit optimism and risk-taking, which can improve growth yet also induce overinvestment. CFOs, as leaders of finance, often display prudence or pessimism, potentially counterbalancing CEO optimism. Cash holdings are central to firms’ risk-bearing capacity, financing flexibility, and value, but excessive cash depresses returns. Traditional explanations of cash policies—trade-off, pecking order, and agency theories—generally assume managerial homogeneity and rationality, overlooking executives’ psychological heterogeneity. Upper echelons theory asserts that firm decisions reflect managerial characteristics, implying that CEO and CFO traits and their interaction may shape cash policies. The paper addresses gaps in prior work that largely examined firm-level factors or single-executive traits, by analyzing the compound effect of CEO optimism and CFO pessimism, and assessing boundary conditions from formal (ownership: SOE vs non-SOE) and informal (regional gambling culture) institutions in China’s transitioning capital market. The authors posit competing hypotheses that the CEO–CFO mix could reduce or increase cash holdings, and expect stronger effects where gambling culture is stronger and among non-SOEs.

Prior studies find CEOs are typically more optimistic and risk-taking than the general population, and CEO optimism can drive investment and financing choices, sometimes leading to overinvestment. CFOs often exert strong influence on financial policies and accounting practices, with greater prudence and conservatism linked to higher cash buffers. Research on CEO–CFO dyads has emphasized observable traits (tenure alignment, social ties, gender) or interactions affecting tax avoidance and M&A outcomes, but has rarely examined how heterogeneous psychological traits across CEO and CFO jointly shape cash policies. On cash holdings, determinants have been analyzed under trade-off, pecking order, and agency frameworks (e.g., uncertainty, diversification, governance), and more recently via managerial characteristics (e.g., gender, optimism). However, the combined effect of CEO optimism with CFO pessimism on cash holdings remains underexplored. The paper builds on upper echelons theory and informal/formal institutional perspectives to study this interaction and its contingencies (regional gambling culture as risk-tolerance proxy; ownership as a channel of political intervention).

Data: Non-financial Chinese A-share listed firms, 2010–2018. Exclusions: financial firms; firms listed after 2010; PT/ST/delisted during sample period; observations with missing values. Final sample: 386 firms, 3,474 firm-year observations. Continuous variables winsorized at 1% and 99%. Data sources: CCER and CSMAR. Variables: Cash holdings measured two ways: Cash1 = (cash + traded financial assets)/total assets; Cash2 = (cash + traded financial assets)/non-cash assets. CEO optimism (Opt): proxy via earnings forecast bias. If (forecast − actual)/actual > 10% (threshold varied to 5%, 15%, 20% in robustness), Opt = 1; else 0. CFO pessimism (Pes): proxy via debt conservatism. Estimate target leverage from a capital structure model (following Marchica and Mura, 2010) using firm fundamentals; if (target − actual)/actual > 10% (also tested 5%, 15%, 20%), define as financially conservative (pessimistic) CFO, Pes = 1; else 0. Controls: Firm-, board-, CEO-, and CFO-level variables per Opler et al. (1999) and related literature; plus industry and year fixed effects. Estimation: OLS baseline (fixed effects deemed unsuitable due to limited within-firm variation in main variables). Core model: Cash_{i,t} = α0 + α1 Pes_{i,t} + α2 Opt_{i,t} + α3 (Pes×Opt){i,t} + Controls + Industry + Year + u{i,t}. Test H1 via α3; then split samples by regional gambling culture and ownership type to test H2 and H3. Contextual measures: Regional gambling culture measured by provincial Welfare Lottery sales/GDP relative to annual provincial average (alternative: accident insurance income/GDP). Ownership: SOE vs non-SOE categories. Robustness and endogeneity checks: (1) Alternative Opt/Pes thresholds (5%, 15%, 20%); (2) Excess cash (residuals from a cash determinants model) as dependent variable; (3) Alternative gambling culture measure; (4) Include financing constraint KZ index; (5) Lagged dependent variable; (6) Beta regression for proportion Cash1; (7) Firm fixed effects; (8) Instrumental variables using regional topographic relief (Rdls) as IV for the dyadic trait combination; (9) Panel stationarity and multicollinearity checks (VIF < 2). Further analyses: Moderators of the CEO–CFO compound effect: (a) CFO board membership (CFO sits on board vs not); (b) CEO–CFO education configurations (both ≥ bachelor’s; both < bachelor’s; CEO ≥ bachelor’s/CFO < bachelor’s; CEO < bachelor’s/CFO ≥ bachelor’s).

• Descriptive statistics: Cash1 range 0.019–0.733; Cash2 range 0.020–2.748, indicating substantial cross-firm variation. Mean Opt = 0.280 (28% optimistic CEOs); mean Pes = 0.419 (41.9% pessimistic CFOs). Correlations: Opt with Cash1/Cash2 = −0.103/−0.089 (p<0.01); Pes with Cash1/Cash2 = 0.181/0.163 (p<0.01). No severe multicollinearity (VIF ≤ 2).
• Baseline compound effect (Table 1): Pes positively associated with cash (Cash1 α≈0.028; Cash2 α≈0.073; both p<0.01). Interaction Opt×Pes significantly negative: Cash1 α≈−0.027 (t≈−3.05), Cash2 α≈−0.084 (t≈−3.31), supporting H1a (the combination reduces cash holdings). Graphical moderation (Fig. 3) confirms mitigation of CFO pessimism’s positive effect by CEO optimism.
• Regional gambling culture (H2): Interaction negative in both subsamples, stronger where gambling culture is strong: Opt×Pes coefficients for strong vs weak regions approximately −0.044 vs −0.018 (Cash1) and −0.163 vs −0.049 (Cash2), indicating a more pronounced cash reduction effect in high-gambling cultures.
• Ownership (H3): Interaction negative for both SOEs and non-SOEs, but larger in magnitude and more stable among non-SOEs than SOEs, consistent with weakened executive trait effects under SOE political intervention.
• Further analyses: (a) CFO board membership: the negative compound effect is weaker when the CFO sits on the board (greater CFO voice softens the CEO-driven reduction in cash); more pronounced when the CFO is not on the board. (b) Education: the negative effect is strongest when both CEO and CFO have at least a bachelor’s and when the CEO has lower education while the CFO has higher education; when both are below bachelor’s, the interaction may turn positive, indicating low education attenuates or reverses the cash-reducing compound effect.
• Robustness: Findings persist under alternative Opt/Pes thresholds (5%, 15%, 20%), with excess cash as dependent variable, alternative gambling culture proxy (accident insurance income/GDP), controlling for KZ financing constraint, lagged cash, beta regression, firm fixed effects, and IV regressions using regional topographic relief. Panel stationarity holds; no reverse causality evident from lag structure.

The research question asks whether optimistic CEOs paired with pessimistic CFOs constitute an effective executive configuration for cash policy. Results indicate that while CFO pessimism alone raises cash buffers (precautionary motive), CEO optimism moderates and reverses this tendency, leading to lower cash holdings when both traits co-occur. This aligns with upper echelons theory: the CEO’s strategic influence and optimism bias reduce precautionary cash hoarding, reflecting lower perceived liquidity risk and greater reliance on internal funds for investment. Contextual factors shape the strength of this effect: stronger regional gambling cultures amplify managerial risk tolerance, exacerbating CEO-led cash reductions and dampening CFO precaution, while SOE governance contexts, with greater political intervention and lower managerial discretion, dilute executive-trait impacts on cash policies. Internal power structures and human capital further condition outcomes: CFO board membership increases the CFO’s influence and weakens the cash-reducing compound effect; higher CFO education relative to the CEO amplifies the negative effect, consistent with greater technical sway on financial policy. Overall, the findings demonstrate how psychological heterogeneity within the CEO–CFO dyad, mediated by institutional environments and governance structures, translates into tangible financial policy choices on cash holdings.

This study shows that firms led by an optimistic CEO and a pessimistic CFO tend to hold less cash, indicating the dyad functions as an effective pairing for improving cash holding efficiency and reducing excess cash. The negative compound effect is stronger in regions with a strong gambling culture and among non-SOEs. Further, internal governance and human capital shape the boundary conditions: the effect weakens when the CFO sits on the board and strengthens when the CFO’s education exceeds the CEO’s or when both are highly educated. Contributions include (1) extending upper echelons theory by modeling compound CEO–CFO psychological traits on financial policy; (2) broadening cash holding determinants to executive trait interactions; and (3) identifying formal and informal institutional contingencies. Future research could refine trait measurement using psychometrics and text analytics, unpack mechanisms in CEO–CFO interactions and stakeholder dynamics, and test generalizability across countries and cultures.

• Measurement: No unified, validated standard for CEO optimism and CFO pessimism; proxies based on earnings forecast bias and leverage conservatism may be noisy despite thresholds and robustness checks.
• Mechanisms: The causal pathways and micro-level interactions within the CEO–CFO dyad are not fully identified; further examination of communication, power dynamics, and stakeholder influences is warranted.
• External validity: Results are drawn from Chinese listed firms; cross-country studies are needed to assess generalizability under different institutional and cultural settings.