Economic policy uncertainty and cash dividend policy: evidence from China

The study examines whether and how economic policy uncertainty (EPU) affects cash dividend policies of Chinese A-share listed firms. Against a backdrop of frequent macro policy adjustments in China (monetary, fiscal, industrial) and the recognized macro- and micro-level effects of EPU, the authors posit that dividend policy—critical for shareholder returns and market stability—may respond systematically to policy uncertainty. Competing forces are outlined: higher EPU can raise operational risk and external financing frictions, potentially discouraging payouts, while it may simultaneously reduce growth opportunities and capital needs, potentially encouraging payouts. Investor demand for safer, current cash flows may also increase with uncertainty, pressuring firms to pay. The paper’s purpose is to identify the net effect of EPU on dividend propensity and scale, explore heterogeneous effects across firm characteristics, uncover mechanisms (investor demand, agency costs, growth opportunities), and assess whether higher payouts during elevated EPU align with shareholder value maximization.

Prior work documents EPU’s adverse effects on investment (e.g., wait-and-see behavior, reduced efficiency) and financing (greater constraints and costs, lower debt supply), as well as effects on cash holdings, M&A, transparency, and reporting. Dividend policy determinants span institutional context, industry environment, and firm characteristics (ownership type, ownership concentration, governance). Building on classic and modern dividend theories (bird-in-hand, irrelevance, tax preference, signaling, agency, life-cycle), the study develops hypotheses: H1: EPU positively affects cash dividends; H2: the positive effect is stronger for small firms; H3: stronger for non-state-owned enterprises (non-SOEs) than SOEs; H4: stronger for firms with low ownership concentration; H5: stronger for firms with low financing constraints.

Data: A-share firms on Shanghai and Shenzhen exchanges, 2007–2021. Exclusions: firms with H/B shares, financial/insurance, ST/ST+ firms, IPO year observations, negative net assets or missing data. All continuous variables winsorized at 1st/99th percentiles. Final sample: 4,421 firms, 38,601 firm-year observations. EPU measure: China EPU from Huang and Luk (2020), annualized by averaging monthly values then dividing by 100. Dividend variables: Payer (dummy=1 if pays cash dividend), Perdiv (cash dividend per share; 0 if none), Payratio (cash dividend per share / EPS). Controls: firm-level (Lev, ROA, Age, Size, EPS, Share1, Board, Indir, Dual) and macro-level (GDPGR, M2g, MESI, CPI). Model: Firm fixed-effects panel regressions of dividend outcomes on EPU and controls. Year fixed effects are not included due to perfect collinearity with EPU; macro controls are included to mitigate omitted time-varying factors. Standard errors clustered at firm level. Endogeneity: Two-stage least squares with an instrumental variable for EPU—US political polarization (Polar), defined as political fractionalization × (1 − government seat share). First-stage relevance and identification confirmed via Kleibergen-Paap LM and Wald F statistics; second-stage examines instrumented EPU effects on dividend outcomes. Robustness: (a) Alternative dividend measures (Divsale, Divy); (b) Alternative EPU measures: December-only EPU (EPUZ2) and Davis et al. (2019) EPUDavis; (c) Event-year dummies for 2008 global financial crisis, 2015 market turmoil, 2020 COVID-19; (d) Semi-mandatory dividend policy controls via refinancing proxies Dumfi (potential refinancing need by above-mean sales growth) and Dumseo (next-year share/rights issues); (e) Two-way clustering by firm and year. Heterogeneity analyses: subsamples by firm size (small vs large), ownership type (non-SOE vs SOE), ownership concentration (low vs high; top-5 shareholders’ combined stake), and financing constraints using SA index (Hadlock-Pierce), splitting at the mean. Mechanism analyses: Two-step approach. Step 1: Regress mechanism variables on EPU. Mechanism variables: investor dividend demand (PDND, Baker-Wurgler dividend premium), agency costs (Fee = administrative expenses / main business income), and growth opportunities (Growth). Step 2: Regress dividend outcome (primarily Perdiv, also Cashdiv as denoted) on predicted mechanism values to assess channels. Further analysis on firm value: Regress MB and Tobin’s Q on EPU, orthogonalized dividend measures (Perdiv_orth, Payratio_orth), their interactions with EPU, and controls, using firm fixed effects. Orthogonalization removes the component of dividend measures explained by EPU and controls from the baseline model.

- Baseline effects (Table 3): EPU positively and significantly increases dividend propensity and scale. With full controls and firm FE: Payer coefficient 0.1048 (t=6.97), Perdiv 0.0228 (t=5.16), Payratio 0.0477 (t=5.10), all at 1% significance. Size is positively related to payouts; leverage is negatively related. Macro controls are generally significant. - Endogeneity (IV; Table 4): Instrument Polar strongly predicts EPU (first-stage coefficient 10.3904, t=470.54; underidentification and weak-ID tests passed). Second-stage shows instrumented EPU significantly increases Payer (0.1136, t=4.87), Perdiv (0.0157, t=2.16), Payratio (0.0492, t=2.95), confirming causality. - Robustness (Table 5): Using Divsale and Divy, EPU remains positive and significant at 1% (e.g., Divsale 0.0044, t=4.18; Divy 0.0019, t=6.25). Alternative EPU measures (EPUZ2, EPUDavis) yield consistent positive coefficients for Perdiv and Payratio. Results persist when controlling 2008/2015/2020 event dummies (Table 6) and semi-mandatory dividend policy proxies Dumfi and Dumseo (Table 7). Two-way clustering by firm and year retains positive significance (Table 8; e.g., Payer 0.0744, t=1.90; Perdiv 0.0151, t=2.29; Payratio 0.0353, t=2.93). - Heterogeneity: • Size (Table 9): Strong positive effect for small firms (Payer 0.0842, t=4.55; Perdiv 0.0246, t=5.37; Payratio 0.0394, t=3.04); insignificant for large firms—supports H2. • Ownership type (Table 10): Significant for non-SOEs (Payer 0.1044, t=5.40; Payratio 0.0629, t=4.73), insignificant for SOEs—supports H3. • Ownership concentration (Table 11): Stronger for low-concentration firms (Payer 0.0922, t=4.58; Perdiv 0.0134, t=3.25; Payratio 0.0483, t=3.85); weaker/insignificant for high concentration—supports H4. • Financing constraints (Table 12): Stronger for low-constraint firms (Payer 0.0596, t=3.34; Perdiv 0.0193, t=3.74; Payratio 0.0212, t=1.72), weaker/insignificant for high-constraint firms—supports H5. - Mechanisms (Table 13): • Investor dividend demand: EPU raises PDND; predicted PDND significantly increases cash dividends (coefficient ~0.0940, t=2.31). • Agency costs: EPU raises Fee (0.0103, t=5.38); higher Fee reduces cash dividends (−0.0345, t=−13.63). • Growth opportunities: EPU lowers Growth (−0.0126, t=−3.34); lower Growth associates with higher cash dividends (−0.5000, t=−2.41 when Growth predicts Cashdiv), indicating a substitution toward payouts when opportunities decline. - Firm value (Table 14): EPU negatively affects MB and Tobin’s Q (e.g., MB −0.8789, t=−15.21; Tobin’s Q −0.9029, t=−13.06). Interactions between EPU and payout measures are positive and significant (e.g., Payer×EPU positive at 1%), indicating that higher payouts during elevated EPU are associated with higher firm value, consistent with shareholder wealth maximization. Overall, results support H1–H5 and are robust across models, measures, and identification strategies.

The findings demonstrate that Chinese listed firms increase both the likelihood and magnitude of cash dividend distributions when economic policy uncertainty rises. This behavior addresses heightened investor demand for near-term cash flows, mitigates agency costs by reducing free cash flow available for potential overinvestment, and reflects reduced marginal value of internal investment when growth opportunities contract under uncertainty. The heterogeneous results align with economic intuition and institutional context: small, non-SOE, low-ownership-concentration, and less financially constrained firms are more sensitive in using dividends to signal stability and cater to investor preferences. The instrumental variables approach and extensive robustness checks bolster causal interpretation. Importantly, the further analysis shows that proactive dividend policies can partially offset the negative valuation impact of EPU, indicating a role for payout policy as a stabilizing mechanism that aligns with shareholder wealth maximization during uncertain policy environments. These results enhance the literature by linking macro policy uncertainty to payout behavior and by clarifying operative channels through investor demand, agency considerations, and investment opportunity sets.

The paper establishes that higher economic policy uncertainty causally increases firms’ propensity to pay and the scale of cash dividends among Chinese A-share listed companies. The effects are stronger for small firms, non-SOEs, firms with dispersed ownership, and firms facing lower financing constraints. Mechanism tests show that EPU raises investor dividend demand and agency costs while lowering growth opportunities, collectively pushing firms toward higher payouts. Further, increasing payouts during heightened EPU is associated with higher firm value, consistent with shareholder wealth maximization. Policy recommendations include enhancing stability and coherence of macroeconomic policies, tailoring support to firm types that are more sensitive to EPU, strengthening governance and delisting mechanisms, and encouraging firms to pair profitability improvements with investor-friendly payout practices and robust risk management. Future research could explore additional mediating mechanisms and employ diversified data sources for measuring EPU.

- EPU measures rely on media-based textual indices, potentially subject to media bias and subjectivity, which may affect accuracy. Combining multiple data sources and methodologies could improve measurement fidelity. - The study focuses on three channels (investor demand, agency costs, growth opportunities); other mediating mechanisms (e.g., corporate governance structures, behavioral factors, product market competition) may further explain the EPU–dividend linkage and merit investigation.