When Policy Becomes Unpredictable, Companies Stop Innovating Alone: AI and ESG Reshape the Rules

Economic policy uncertainty is often treated as an obstacle to technological progress, forcing businesses to postpone investment and reconsider ambitious research programmes. A major study of Chinese companies reveals a more complicated response: while uncertainty is associated with declining independent innovation, it coincides with greater collaboration between firms, suggesting that unpredictable policy environments can change how technologies are developed rather than simply discouraging development.

Published in the journal Systems, the study, titled "Firm-Perceived Economic Policy Uncertainty and Open Innovation, as Well as the Role of AI-Related Asset Intensity and ESG Performance," was conducted by Kunfeng Chen, Zaisheng Zhang, Meng Liu and Xiangchun Li. Drawing on 4,828 publicly listed Chinese companies between 2010 and 2024, the researchers identify artificial intelligence-related assets, corporate sustainability performance and institutional ownership as important factors associated with collaborative innovation under uncertainty.

The findings challenge conventional interpretations of corporate risk aversion. Instead of treating innovation as a choice between investing and waiting, the research shows that businesses may reorganize technological development around partnerships, shared resources and external knowledge. However, the advantages appear concentrated among larger enterprises, raising questions about whether smaller companies possess sufficient capabilities to respond similarly.

Uncertainty Is Not Killing Innovation. It Is Changing Who Innovates Together

Corporate innovation involves substantial commitments of capital, personnel and technological resources. When government policies become difficult to predict, companies face greater uncertainty about future demand, regulatory requirements and potential investment returns. Delaying expensive research projects can therefore appear financially prudent, particularly when development costs cannot easily be recovered.

Economic research has long produced conflicting conclusions about these conditions. Some studies associate policy uncertainty with weaker innovation, while others find that competitive pressure encourages companies to pursue technological opportunities despite unpredictable conditions. The researchers suggest that these disagreements partly reflect insufficient attention to how businesses organize their innovation activities.

Their investigation distinguishes independent patent applications from jointly filed patents, using the latter as an observable measure of collaborative open innovation. The analysis covers 41,283 firm-year observations, combining patent records, corporate financial information and annual-report disclosures to examine how companies respond to perceived policy uncertainty.

The central finding is striking. Higher firm-perceived economic policy uncertainty was significantly associated with fewer independent patent applications but more jointly filed patents. The difference between these relationships was statistically significant, indicating that uncertainty relates differently to innovation undertaken internally and technological development pursued through cooperation.

The positive association extended across invention, utility-model and design patents. In an alternative statistical model using patent counts, a one-standard-deviation increase in perceived uncertainty was associated with approximately 17.98% more joint patent applications and 18.89% more joint invention applications in the relevant estimation samples.

These estimates establish associations rather than causal effects. Nevertheless, they support the researchers' argument that firms facing uncertainty may preserve technological opportunities through partnerships that provide complementary expertise and distribute research commitments.

Collaborative innovation also introduces additional coordination requirements and risks surrounding knowledge sharing. Its attractiveness therefore depends on whether companies can establish reliable partnerships and manage the organizational complexities involved.

AI Emerges as a Link Between Policy Risk and Collaborative Research

The study identifies AI-related asset intensity as an important channel connecting perceived policy uncertainty with collaborative innovation. Companies expressing greater uncertainty tended to hold more AI-related assets relative to their overall asset base, while those assets were positively associated with joint patenting.

Statistical mediation analysis estimated that AI-related asset intensity accounted for approximately 8.45% of the overall relationship between perceived uncertainty and collaborative innovation. The corresponding proportion for jointly filed invention patents reached 9.45%, indicating a statistically meaningful but partial contribution.

The researchers interpret this relationship through growing information-processing demands. Policy uncertainty complicates assessments of technological opportunities, potential collaborators and future market conditions. AI-related resources may support information analysis, external knowledge searches and coordination between organizations pursuing common research objectives.

Such capabilities could help companies identify complementary expertise and revise innovation decisions as new information emerges. Rather than relying exclusively on established internal research systems, firms may use technological infrastructure to evaluate opportunities across organizational boundaries.

An important qualification concerns how the study measures AI. Its indicator captures recorded AI-related software and hardware assets, including some broader digital infrastructure, as a proportion of corporate assets. It does not directly measure the deployment, effectiveness or sophistication of AI applications.

Consequently, the findings do not establish that particular AI systems independently generated collaborative inventions. They instead identify a relationship consistent with technological resource allocation supporting external innovation.

Purchasing technology does not automatically improve collaboration; the researchers emphasize that effective utilization requires appropriate data, skilled personnel and integration into organizational processes.

ESG and Investor Oversight Become Part of the Innovation Equation

In addition to technological resources, the research examines whether environmental, social and governance performance influences companies' willingness and ability to innovate collaboratively.

The findings indicate that stronger overall ESG performance reinforces the positive relationship between perceived policy uncertainty and jointly filed patents. Environmental, social and governance scores each demonstrated positive moderating associations, although their statistical strengths varied across innovation categories.

The proposed explanation centers on confidence between collaborating organizations. Joint research requires participants to exchange knowledge, contribute resources and sustain commitments despite uncertain commercial or regulatory conditions. Strong ESG performance may provide information that helps potential partners evaluate corporate reliability and longer-term organizational conduct.

Investor ownership introduces another dimension. The study finds that higher institutional ownership strengthens ESG's moderating relationship with collaborative innovation, suggesting that external oversight may influence how corporate governance supports cooperation.

Institutional investors can potentially reinforce accountability by monitoring management decisions and corporate commitments. In the researchers' framework, such oversight complements the trust and continuity needed for partnerships involving shared technological development.

However, the analysis does not directly observe trust between innovation partners or measure how effectively institutional investors monitor individual collaborations. ESG scores and ownership structures therefore serve as indicators associated with these proposed governance mechanisms, not definitive proof that they operate as described.

The research nevertheless expands the discussion of ESG beyond sustainability reporting and investment performance. It suggests that governance characteristics may be relevant to the organizational conditions under which companies pursue joint innovation during uncertain periods.

For businesses, this points toward integrating credible governance practices into partnership selection and collaborative project management while retaining contractual safeguards.

The Size Divide Exposes Who Can Innovate Through Uncertainty

The positive association between perceived policy uncertainty and collaborative patenting was concentrated among larger companies, while corresponding relationships among smaller firms were not statistically significant.

Formal comparisons confirmed significantly stronger relationships among larger enterprises for both overall collaborative innovation and joint invention patents. The researchers attribute this pattern to potential differences in financial capacity, technological resources and organizational capabilities required to identify partners and coordinate complex research projects.

Corporate ownership also shaped the results. Both state-owned and non-state-owned firms displayed positive associations between uncertainty and collaborative innovation, but the relationship was significantly stronger among state-owned enterprises specifically for invention-type collaborative patenting.

The researchers suggest that state-owned companies' institutional connections and access to resource-coordination networks may help explain their stronger response in this category. However, the evidence does not establish a significant ownership-group difference for overall collaborative patenting.

These differences complicate the assumption that collaborative innovation is equally accessible to all businesses. Partnerships can distribute research costs, but forming and maintaining them still requires resources, organizational expertise and the capacity to integrate knowledge from outside the company.

For innovation policy, the findings support closer examination of barriers that prevent smaller enterprises from participating in external research networks. Measures facilitating knowledge exchange and cooperation may be relevant, although the study does not evaluate specific policy interventions.

The evidence also has geographical boundaries. Researchers examined Chinese A-share listed companies, excluding several sectors and observations without required information. Their findings cannot automatically be extended to unlisted businesses or companies operating under different institutional arrangements.

Methodological safeguards strengthen the analysis, including alternative uncertainty indicators, different patent definitions, lagged variables and additional statistical tests. Yet joint patents capture only part of open innovation, leaving licensing, informal knowledge exchanges and collaborations without jointly filed patents largely unobserved.

Annual-report language may also reflect managerial disclosure practices rather than policy exposure alone. Similarly, the analysis cannot fully separate cooperation between independent organizations from collaboration within corporate groups.

More research combining patent records, detailed ownership information and organizational surveys could clarify these distinctions. Direct examination of AI deployment, partner relationships and innovation outcomes would also help establish how technological resources and governance practices influence cooperation.