Cultural inclusion and corporate sustainability: evidence from food culture and corporate total factor productivity in China

The study examines whether and how regional cultural inclusion affects corporate sustainability, operationalized as firms’ total factor productivity (TFP). Motivated by growing evidence that culture influences financial systems and firm behavior, the authors focus on inclusion as a geographic cultural characteristic reflecting local acceptance and respect for external cultures. Using China as a setting—given its strong geocultural norms, diverse regional cultures, and significant internal migration—the paper asks whether higher cultural inclusion enhances firms’ sustainability and through what mechanisms. The authors posit competing hypotheses: inclusion may promote sustainability by improving external environments and internal governance (Assumption 1a), or hinder it through cultural conflicts and knowledge integration barriers (Assumption 1b). They also hypothesize that CEOs’ birthplace inclusion strengthens the link between local inclusion and firm sustainability (Assumption 2).

Prior work links culture to economic and financial outcomes, including corporate governance and behavior (Guiso et al., 2006, 2009; Chen et al., 2014). Organizational inclusion affects business development (Shore et al., 2011, 2018), yet the impact of external, regional inclusion on firms is underexplored. Studies have proxied geoculture with dialect diversity (Lei et al., 2022), arguing language encapsulates cultural patterns. The paper extends this by using food culture as a proxy for cultural inclusion, grounded in literature that food reflects natural geography, social bonds, and local identity (Cook, 2006, 2008; Cook et al., 2011; Cappellini and Yen, 2013; Lo Monaco and Bonetto, 2019). Food taste symbols (e.g., spiciness in Sichuan/Hunan, umami in Guangdong/Fujian) embody regional identity (Chang et al., 2010; Li et al., 2021; Trachootham et al., 2018). Cultural inclusion could shape knowledge valuation and commercialization (Audretsch et al., 2010; Desrochers and Leppala, 2011), alleviate financing constraints (Campello et al., 2010), and influence internal diversity, humanism, and governance (Van Knippenberg and Schippers, 2007; Qian, 2013; Kim et al., 2011). The paper contributes to culture–finance and TFP literatures that have emphasized formal institutions and firm characteristics (Olley and Pakes, 1996; Petrin, 2003; Hsieh and Klenow, 2009) by introducing informal cultural inclusion as an antecedent of firm TFP.

Setting and sample: A-share Chinese listed firms from 2010–2019 (excluding ST/PT, financial and real estate), yielding 10,140 firm-year observations for 2,118 firms. Financials and executives from CSMAR; regional socioeconomics from the National Bureau of Statistics of China. Cultural inclusion construction: 1) Recipe data (Aug 2017) were scraped from Meishijie, classifying cuisines and quantifying each cuisine’s seven taste dimensions: spicy, sweet, salty, sour, umami, pungent, oily. Ingredients were mapped via text analysis, allowing multi-taste attribution (e.g., ketchup labeled sour, sweet, oily). Cuisine-level taste vectors were computed by equally averaging dishes. 2) Restaurant point-of-interest (position of catering, PC) data from Amap (over 7 million entries, 383,321 valid PCs identified with 16 cuisines) were used to estimate each province’s cuisine mix. Provincial taste vectors were computed as weighted averages of cuisine taste vectors using provincial PC shares. Cultural inclusion index (C_Inclu_p): For each province p, the absolute relative deviation between the provincial taste vector and its corresponding native cuisine taste vector was computed for each taste, then averaged over seven tastes: D_taste_p = |taste_pr − taste_diet_p| / taste_diet_p; C_Inclu_p = average of seven D_taste_p. Alternative inclusion proxies: Out_cui (share of non-local cuisines in a province) and Dspicy (spicy dimension deviation). Dependent variable: Firm total factor productivity via Levinsohn–Petrin semiparametric approach, log-transformed (TFP_LP). Robustness uses Olley–Pakes TFP (TFP_OP). Controls: Lagged firm-level variables (Size, LEV, BM, RET, Attention, Turnover, ROA, Top_10, Top_1, Lncom, Boardsize, Indeboard, Separation, CO_CEO, SOE) and regional variables (GDP_per, GDP_growth, Pop_growth, Consume_per), with industry and year fixed effects; robust clustered standard errors at firm and year levels. Baseline model: TFP_LP_ipt = α + β C_Inclu_p + γ Controls_{i,p,t-1} + FE + ε. Endogeneity strategies: • Instrumental variables (2SLS) using mountainous undulation (Moun_Undu; Feng et al., 2008) as an instrument for inclusion (first stage: inclusion on Moun_Undu and controls; second stage: TFP_LP on fitted inclusion and controls). • Regression discontinuity design (RDD) at provincial borders with similar cultural baselines but differing inclusion: Shanghai vs. Jiangsu (RDD_SJ) and Shanghai vs. Zhejiang (RDD_SZ); indicator equals 1 for Shanghai, 0 for neighbor; includes fixed effects as specified. Mechanism test (humanistic care): Two-stage approach using CSR_staff (CSR toward employees from Hexun) as mediator. First stage: CSR_staff on inclusion and controls; second stage: TFP_LP on fitted CSR_staff and controls. Transmission tests: • CEO birthplace inclusion (Birth_Inclu) regressed on TFP_LP to test whether executives’ home-region inclusion predicts firm sustainability. • More_Inclu dummy equals 1 if CEO’s birthplace inclusion exceeds the firm’s current location inclusion; tested for amplification effects. Heterogeneity analyses: Interactions with competition (Com based on industry HHI median; C_Inclu_Com interaction) and analyst attention (atten based on median; C_Inclu_Atten interaction).

• Cultural inclusion positively associates with firm sustainability (TFP): In baseline regressions, C_Inclu coefficients are positive and significant at 1%: e.g., 0.140 (t=8.253) without regional controls; 0.122 (t=5.229) with controls. Using Out_cui yields similar effects: 0.158 (t=5.802) and 0.155 (t=5.127). • Instrumental variables (2SLS) confirm causality: First stage shows Moun_Undu strongly and negatively predicts inclusion (e.g., −0.129, t=−45.240). Second stage: fitted C_Inclu significantly increases TFP_LP (e.g., 0.352, t=6.131). Weak-instrument tests are strong (Cragg-Donald F≈2360; Kleibergen-Paap F≈2047). Using Out_cui as the endogenous regressor provides consistent results. • Regression discontinuity: Firms in Shanghai exhibit significantly higher TFP_LP relative to Jiangsu (RDD_SJ ≈ 0.111–0.103, both p<0.01) and Zhejiang (RDD_SZ ≈ 0.086–0.116, p<0.01), consistent with higher inclusion driving sustainability. • Robustness to inclusion measures: Using Dspicy as the inclusion proxy yields positive, significant coefficients (≈0.019–0.021, p<0.01). • Robustness to TFP measure: With TFP_OP, inclusion measures remain positive and significant (e.g., C_Inclu ≈ 0.093–0.104, p<0.01; Out_cui ≈ 0.224–0.225, p<0.01). • Transmission via executives: CEO birthplace inclusion (Birth_Inclu) predicts higher TFP_LP (≈0.080–0.104; p<0.05 to p<0.01). More_Inclu (CEO from more inclusive region than firm location) strengthens the inclusion–TFP link (positive and significant). • Mechanism (humanistic care): Inclusion increases CSR toward staff (CSR_staff) (e.g., C_Inclu → CSR_staff: 0.589, t=4.145; Out_cui → CSR_staff: 0.448, t=2.429; Dspicy → CSR_staff: 0.091, t=3.954). Fitted CSR_staff significantly raises TFP_LP (e.g., 0.357–0.381, p<0.01), supporting the humanistic care hypothesis. • Heterogeneity: The positive effect of inclusion on TFP is stronger in more competitive industries (interaction C_Inclu_Com ≈ 0.059–0.063, p<0.10) and among firms with higher analyst attention (C_Inclu_Atten ≈ 0.063, p<0.01). • Descriptive statistics highlight substantial cross-province variation in inclusion (C_Inclu mean 0.686; SD 0.319; min 0.065; max 1.191) and firm TFP.

The results support Assumption 1a: regional cultural inclusion enhances corporate sustainability by fostering a more supportive external environment and by shaping more humanistic, tolerant internal governance that benefits employees and organizational cohesion. The IV and RDD strategies mitigate endogeneity and bidirectional causality, reinforcing a causal interpretation from inclusion to higher TFP. The mechanism analysis indicates that inclusion encourages firms to increase CSR toward employees, which in turn boosts productivity, aligning with theories linking worker well-being and organizational citizenship to performance. Executive transmission results show that leaders import cultural values from their birthplaces into corporate decisions, amplifying local inclusion effects. The heterogeneity findings suggest that inclusion’s benefits are especially pronounced when competitive pressures are high and when external monitoring via analysts is stronger, consistent with inclusion-driven humanistic practices translating into competitive advantages and reputational benefits.

The paper introduces and operationalizes a regional cultural inclusion index using food taste deviations and demonstrates that firms in more culturally inclusive regions achieve higher TFP, indicating stronger sustainability. Causal evidence is provided via an instrumental variable based on mountainous undulation and border RDDs. Inclusion’s effect operates partly through greater humanistic care for employees, and CEOs transmit cultural inclusion from their birthplaces into firm practices. The study broadens the culture–finance literature by highlighting informal institutions as determinants of firm productivity and sustainability. Future research directions include constructing time-varying inclusion measures, expanding analyses beyond China to test for non-linear effects (e.g., potential downsides of excessive inclusion), and addressing cross-country endogeneity to generalize the findings.

• The cultural inclusion index is constructed from a single year of restaurant and recipe data, limiting temporal variation. • The study focuses on China; while offering control over macro-institutional differences, it constrains external validity and precludes testing whether excessive inclusion could harm sustainability. • Cross-national extensions face strong endogeneity challenges that require careful identification strategies.