Enterprise digital transformation, managerial myopia and cost stickiness

The study examines whether and how enterprise digital transformation reduces cost stickiness, an asymmetric cost behavior where costs increase more with rising activity than they decrease with falling activity. In the digital economy era, firms adopt big data, AI, blockchain, and cloud technologies that can improve information flows, decision-making, and monitoring, potentially lowering adjustment and financing costs that underlie cost stickiness. The paper posits that digital transformation can alleviate internal information frictions and external agency and financing frictions, thereby reducing sticky costs. It also investigates the moderating role of managerial myopia, which may raise adjustment costs and financing constraints and thus undermine digital transformation’s benefits for cost stickiness. The importance lies in improving cost management, forecasting accuracy, firm performance, and risk by mitigating cost stickiness.

Prior research shows digital transformation reshapes processes, business models, and value creation, improving performance, productivity, innovation, CSR, and competitive advantage. Cost stickiness reflects asymmetric resource adjustment driven by adjustment costs, managerial optimism, and agency problems, contradicting traditional symmetric cost models. Empirical evidence documents cost stickiness across countries and links higher adjustment costs and agency issues to greater stickiness. Emerging evidence suggests digital technologies and Internet development can reduce cost stickiness, and financing constraints exacerbate it. This paper extends literature by: (1) evaluating both operating and SG&A cost stickiness (with labor cost as robustness), (2) integrating internal (adjustment costs) and external (financing constraints) mechanisms, and (3) introducing managerial myopia as a moderator in the digital transformation–cost stickiness relationship.

Data: Chinese A-share listed companies (Shanghai and Shenzhen) from 2010–2021 sourced from CSMAR. Exclusions: ST/*ST firms, financial firms, firms with <1 year history, and missing data. Final sample: 26,134 firm-year observations (varies by model). Dependent variables: changes in operating costs (ΔlnOC) and SG&A (ΔlnSGA) to measure cost stickiness per Weiss (2010) specification: cost change on revenue change with interaction D×ΔlnReven (D=1 if revenue declines) capturing asymmetry (stickiness if coefficient on D×ΔlnReven is significantly negative). Independent variables: ΔlnReven, D, and Digital (digital transformation). Digital measurement: machine-learning/text analytics counting 216 digital-related keywords (e.g., internet, AI, big data, blockchain) in annual reports; log-transformed. Controls: economic variables (human capital intensity EI; asset intensity AI; city GDP growth GDPG) and their interactions with ΔlnReven; firm-level controls (Age, Size, Lev, Dual, Shrcr1, Big4, Audittyp). Mechanism variables: Adjustment costs (AC) = (Net fixed assets + Construction in progress + Intangibles + Long-term prepaid expenses) / Total assets (Williamson, 1988 proxy for asset specificity/adjustment costs). Financing constraints (SA) index (Hadlock & Pierce, 2010): SA_it = −0.737 Size + 0.043 Size^2 − 0.04 Age (higher indicates tighter constraints). Moderator: Managerial myopia measured by frequency of 43 myopia-related keywords in annual reports (per Guo et al., 2023), scaled by total words×100. Models: Multi-dimensional fixed effects regressions with firm, industry, and year fixed effects and firm-clustered SEs. (1) Baseline stickiness and Digital moderation: Y_it on ΔlnReven, D×ΔlnReven, and D×ΔlnReven×Digital. (2) Mechanism first stage: M_it on Digital. (3) Mechanism second stage: Y_it on ΔlnReven, D×ΔlnReven, D×ΔlnReven×Digital, Digital, and M_it (AC or SA). (4) Moderation by myopia: include Digital, Myopia, and D×ΔlnReven×Digital×Myopia. Robustness and endogeneity: alternative Digital measures (lagged DigitalB, LDigital), alternative dependent variable (labor cost stickiness based on cash paid to and on behalf of employees), PSM, Heckman selection, and IV approaches; extensive FE and controls; Sobel and bootstrap tests (1,000 resamples) for mediation.

Baseline: ΔlnReven strongly predicts cost changes (operating: ~0.982–0.986; SG&A: ~0.551–0.561, all p<0.01). D×ΔlnReven is significantly negative across specifications (e.g., operating: −0.122 to −0.124, SG&A: −0.414 to −0.531), confirming cost stickiness. The interaction D×ΔlnReven×Digital is significantly positive (operating: ~0.020–0.024, p≤0.10; SG&A: ~0.054–0.065, p≤0.05), indicating digital transformation attenuates cost stickiness. Robustness: Results hold using lagged Digital (DigitalB) or alternative digital proxy (LDigital), and when measuring labor cost stickiness. PSM, Heckman, and IV analyses corroborate the causal interpretation; instruments are strong (e.g., high Kleibergen-Paap statistics). Mechanisms: Digital significantly reduces AC (Digital→AC: −0.007, t≈−5.15) and alleviates SA (Digital→SA: −0.003, t≈−2.39). Lower AC and SA reduce stickiness: D×ΔlnReven×AC negatively associated with cost changes (operating: −0.110, SG&A: −0.206), and D×ΔlnReven×SA negative (operating: −0.015, SG&A: −0.106), supporting mediation. Sobel tests significant at 1% and bootstrap CIs exclude zero, confirming both mechanisms. Managerial myopia: Myopia increases AC and SA (positive coefficients) and weakens the Digital effect on reducing stickiness (D×ΔlnReven×Digital×Myopia positive: operating ~0.065, SG&A ~0.146). Heterogeneity: Stronger inhibitory effect in mature-stage firms; effects are weaker/insignificant in growth and decline stages (growth may even see exacerbation for operating costs). Effect is stronger in regions with better digital economy policy environments. By ownership, effects are stronger for state-owned enterprises than non-state-owned. By technology intensity, effects are stronger for high-tech firms than non-high-tech.

Findings confirm that digital transformation mitigates cost stickiness by improving information flows and monitoring, which lowers internal adjustment costs and eases external financing constraints. This addresses the central question by demonstrating that digitalization aligns cost adjustments more symmetrically with revenue changes. However, managerial myopia undermines these benefits by fostering short-termism, empire building, and higher adjustment and financing frictions, thereby attenuating the positive impact of digitalization on cost behavior. The results emphasize the interplay between technology adoption and managerial behavior, and the importance of context: mature, state-owned, high-tech firms and those in supportive policy environments benefit more from digitalization in cost management.

Using 2010–2021 data on Chinese A-share listed firms, the study shows that enterprise digital transformation significantly inhibits cost stickiness for both operating and SG&A costs. Digitalization reduces stickiness via two mechanisms: lowering adjustment costs and alleviating financing constraints. Managerial myopia weakens this beneficial effect by increasing adjustment costs and financing constraints. The impact varies by life cycle stage (strongest in maturity), ownership (stronger for SOEs), technology intensity (stronger for high-tech firms), and regional policy environment (stronger where digital economy policies are better). Robustness and endogeneity checks support the validity of these findings. Policy and managerial implications include promoting digital transformation, tailoring incentives and subsidies to firm characteristics, enhancing policy environments, improving information transparency, easing financing constraints, and designing governance and incentives to curb managerial myopia. Future research should explore additional cost categories and further managerial traits (e.g., CEO overconfidence, narcissism, arrogance) in the digitalization–cost behavior nexus.

The study primarily analyzes operating and SG&A cost stickiness, using labor cost stickiness as a robustness check; broader cost categories warrant further investigation. Managerial characteristics beyond myopia (e.g., CEO overconfidence, narcissism, arrogance) are not examined but may shape the digitalization–cost stickiness relationship. While extensive robustness and endogeneity tests are conducted, unobserved factors or measurement limitations in text-based digitalization and myopia proxies may remain.