Does hometown connection between chairmen and CEOs improve compensation-performance sensitivity in China?

The study investigates whether and how hometown (fellow-villager) connections between chairmen and CEOs affect CEO compensation-performance sensitivity in Chinese private listed firms. In China’s relationship-oriented culture, the laoxiang (fellow villager) identity fosters strong social networks emphasizing trust and reputation, which can influence corporate governance. Two competing mechanisms are posited: a reputation mechanism that enhances alignment and contract enforcement versus a collusion mechanism that may weaken oversight. Focusing on private A-share firms (2011–2020) where compensation practices are market-driven (unlike SOEs), the study tests whether such connections increase pay-performance sensitivity and how this effect varies with regional hometown culture strength and degree of marketization.

Grounded in social identity theory and agency theory, the paper argues that fellow-villager ties form cohesive in-groups that value reputation, facilitating trust and contract enforcement, which can raise CEO pay-performance sensitivity. Conversely, such ties can promote collusion, reduce board independence, and weaken sensitivity. The paper develops competing hypotheses: H1a (reputation mechanism) predicts a positive effect of chairman–CEO fellow-villager ties on pay-performance sensitivity; H1b (collusion mechanism) predicts a negative effect. It further posits contextual moderators: H2 predicts stronger positive effects in regions with stronger hometown culture (higher societal and in-group collectivism per Minkov and Blagoev, 2012). H3 predicts stronger positive effects in regions with higher marketization, where formal institutions and transparency strengthen reputation effects and curb collusion. Prior literature on networks and governance (e.g., Hwang and Kim, 2009; Nguyen, 2012; Kramarz and Thesmar, 2013; Xu et al., 2021) provides mixed expectations, motivating empirical tests.

Sample: Private Chinese A-share firms listed in Shanghai and Shenzhen from 2011–2020. Exclusions: SOEs (administrative pay setting), financial/insurance firms, ST firms, cases where chairman=CEO, firms lacking executive information. Final sample: 1,293 firms, 6,478 firm-year observations (unbalanced panel). Industry distribution spans equipment manufacturing (34.8%), materials/chemical/healthcare (21.5%), etc. Data: CEO remuneration and firm financials from CSMAR. Executive demographics (gender, age, education, birthplace) manually collected from CSMAR profiles, Chinese web sources, and annual reports. Continuous variables winsorized to mitigate outliers. Estimation strategy: Baseline pooled OLS with industry and year fixed effects. Firm fixed effects not used because the key variable (fellow-villager tie) has limited within-firm time variation. Model: Lnpay = β0 + β1 Roe + β2 RoeCountrymen + β3 Countrymen + Controls + Industry FE + Year FE + ε. Lnpay is log CEO total monetary compensation. Roe is return on equity; robustness uses Roa. Countrymen=1 if chairman and CEO share the same province (or city when available), else 0. The interaction term RoeCountrymen captures how hometown ties affect pay-performance sensitivity. Controls: firm size (log assets), leverage, growth, ownership concentration (Top1), proportion of independent directors, remuneration committee dummy, region dummy (eastern vs central/western), plus industry and year effects. Contextual moderators: Regional culture (Reculture=1 if province’s societal+in-group collectivism score exceeds mean) and marketization (Index=1 if province’s marketization index exceeds annual mean) used for split-sample analyses. Endogeneity handling: Exclude family-related chairman–CEO pairs; use instrumental variable (IV) approach with topographic relief (Qfdu) of CEO’s hometown as instrument for the fellow-villager tie (relief lowers population density and thus match probability; unrelated to CEO ability/compensation). Two-stage regressions estimated; first stage shows strong, negative relation between relief and Countrymen; second stage tests effect on sensitivity. Further tests: Moderation by CEO tenure (First-year dummy and continuous Tenure) and by appointment method (Entele=1 internal promotion, 0 external hire) via triple interactions.

• Baseline grouping: Pay–performance sensitivity (Roe coefficient) is higher when chairman and CEO are fellow villagers: 1.086 (Countrymen=1) vs 0.706 (Countrymen=0); difference significant (chi2 p=0.0265).
• Interaction model: RoeCountrymen positive and significant, indicating fellow-villager ties increase pay–performance sensitivity. Without controls: 0.343**; with controls: 0.286 (Table 6).
• Context heterogeneity:
  • Strong hometown culture regions (Reculture=1): Roe*Countrymen = 0.389**; weak culture (Reculture=0): positive but insignificant (Table 7).
  • High marketization (Index=1): RoeCountrymen = 0.331; low marketization (Index=0): positive but insignificant (Table 7).
• Excluding family relations: RoeCountrymen remains positive and significant (0.275** without controls; 0.252 with controls) (Table 8).
• Instrumental variable results: First stage—Qfdu negatively predicts Countrymen (-0.372***). Second stage—RoeCountrymen = 0.136**; F-statistic=24.98, rejecting weak IV (Table 9).
• Robustness checks:
  • Using Roa: RoaCountrymen = 0.564 (Table 10, col. 1).
  • Alternative measurement of ties: province+city definition RoeCountrymen = 0.269 (full sample); city-only definition RoeCountrymen = 1.023** (n=1,839), stronger effect at narrower geographic scope; difference test p=0.0268 (Table 10, cols. 2–3).
  • Bootstrap (1,000 resamples): RoeCountrymen = 0.286, consistent with baseline (Table 10, col. 4).
• Mechanisms:
  • Tenure moderation: First-year CEOs show stronger effect (RoeCountrymenFirst = 0.839**, Table 11). With continuous tenure, RoeCountrymenTenure = -0.081* (effect weakens as tenure increases).
  • Appointment method: RoeCountrymenEntele = -0.669** (Table 12), indicating stronger effect for externally hired CEOs (Entele=0) than internally promoted CEOs (Entele=1).

Findings support the reputation mechanism: chairman–CEO hometown connections increase the alignment and enforceability of CEO pay contracts, making compensation more sensitive to performance. The effect is contingent on institutional and cultural contexts—stronger in provinces with high collectivism (hometown culture) and higher marketization—where reputation costs are higher and formal institutions curb collusion. The diminishing effect with longer CEO tenure and the stronger effect for externally hired CEOs suggest that reputation within the fellow-villager network is especially salient early in relationships and when prior familiarity is lower, while prolonged familiarity can foster connivance that attenuates the reputation channel. Overall, informal ties complement formal governance mechanisms to enhance pay–performance sensitivity but do not substitute for them; formal institutional quality remains crucial.

The study shows that chairman–CEO hometown (fellow-villager) connections in Chinese private listed firms increase CEO compensation–performance sensitivity, particularly in regions with strong hometown culture and higher marketization. Mechanism analyses indicate the effect weakens as CEO tenure increases and is more pronounced for externally hired CEOs than for internally promoted ones. Results are robust to alternative performance measures (Roa), alternative hometown-tie operationalizations (province+city and city-only), bootstrap sampling, exclusion of family firms, and an IV approach addressing endogeneity via topographic relief. The findings highlight the complementary role of informal social relationships with formal governance in shaping effective compensation contracts. Future research could extend to other informal ties, sectors (e.g., SOEs), and cross-country settings, and explore finer measures of executive ability and social capital.

• The study omits direct measures of CEO ability due to measurement difficulty; endogeneity is addressed using an instrumental variable (topographic relief), but residual concerns may remain.
• Hometown connections are primarily measured at the province level when city-level data are unavailable, which may introduce measurement error; robustness tests with city-level data partly mitigate this.
• Sample is restricted to private Chinese A-share firms (2011–2020), excluding SOEs, financial/insurance, ST firms, and cases lacking executive information, which may limit generalizability.
• Some hometown data are missing; randomness was assessed via bootstrap to alleviate concerns.
• Contextual moderators (regional culture and marketization) are based on external indices, which may not capture all nuances of local institutions.