Scholarlink

<p>Policy-led innovation networks are often expected to expand collaboration and improve knowledge recombination, but it remains unclear whether the processes that generate network growth also support more open collaboration structures and citation-efficient impact. To address this issue, this study develops a coupled multi-mechanism framework linking tie formation, emergent network positions, and innovation outcomes in the Guangdong-Hong Kong-Macao Greater Bay Area (GBA). Using 57,846 collaboration records among 3,582 organizations (2009–2023), we estimate phase-segmented exponential random graph models (ERGMs) across three policy stages to jointly assess endogenous structural mechanisms (preferential attachment, triadic closure, and brokerage-related constraints) and exogenous selection mechanisms (organizational type homophily, experience-based cumulative advantage, and policy-recognized elite selectivity, measured using external designation records), and then use negative binomial models to relate network positions to collaborative patenting and forward-citation outcomes. The results reveal a paradoxical trajectory: despite rapid expansion, baseline tie propensity becomes more negative, indicating heightened partner selectivity, while the network consolidates into an “elite-driven community closure” configuration. Preferential attachment remains persistent, triadic closure strengthens over time, and brokerage-oriented configurations are consistently suppressed, with only partial attenuation. Type homophily persists across stages, and elite dynamics become increasingly stratified—elite-elite assortativity emerges in later phases while elites are not universally more active in tie formation—consistent with selective clubbing. Experience effects remain positive but decline over time. Linking network positions to outcomes using forward citations, we observe a pattern consistent with a quantity-quality tension: centrality and closure primarily scale collaborative output and aggregate citation impact, whereas brokerage capacity is more consistently related to citation-efficient impact. These findings suggest that policy-led expansion can coexist with structural inertia when endogenous dependencies and policy-mediated organizing arrangements jointly channel collaboration toward recognized cores. The study points to targeted policy levers: reducing cross-type transaction costs, counteracting hub concentration through peripheral engagement, and building cross-jurisdictional reputation-sharing infrastructure to support credible bridging.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Policy-Mediated Collaboration or Structural Inertia? Elite-Driven Community Closure and the Quantity-Quality Tension in the Greater Bay Area Innovation Network

  • 期刊论文

  • 发表日期：

    2026年7月20日

查看全文
  • Xinyu Guo,
  • Sumin Feng,
  • Rende Li,
  • Liangrong Song

摘要

Policy-led innovation networks are often expected to expand collaboration and improve knowledge recombination, but it remains unclear whether the processes that generate network growth also support more open collaboration structures and citation-efficient impact. To address this issue, this study develops a coupled multi-mechanism framework linking tie formation, emergent network positions, and innovation outcomes in the Guangdong-Hong Kong-Macao Greater Bay Area (GBA). Using 57,846 collaboration records among 3,582 organizations (2009–2023), we estimate phase-segmented exponential random graph models (ERGMs) across three policy stages to jointly assess endogenous structural mechanisms (preferential attachment, triadic closure, and brokerage-related constraints) and exogenous selection mechanisms (organizational type homophily, experience-based cumulative advantage, and policy-recognized elite selectivity, measured using external designation records), and then use negative binomial models to relate network positions to collaborative patenting and forward-citation outcomes. The results reveal a paradoxical trajectory: despite rapid expansion, baseline tie propensity becomes more negative, indicating heightened partner selectivity, while the network consolidates into an “elite-driven community closure” configuration. Preferential attachment remains persistent, triadic closure strengthens over time, and brokerage-oriented configurations are consistently suppressed, with only partial attenuation. Type homophily persists across stages, and elite dynamics become increasingly stratified—elite-elite assortativity emerges in later phases while elites are not universally more active in tie formation—consistent with selective clubbing. Experience effects remain positive but decline over time. Linking network positions to outcomes using forward citations, we observe a pattern consistent with a quantity-quality tension: centrality and closure primarily scale collaborative output and aggregate citation impact, whereas brokerage capacity is more consistently related to citation-efficient impact. These findings suggest that policy-led expansion can coexist with structural inertia when endogenous dependencies and policy-mediated organizing arrangements jointly channel collaboration toward recognized cores. The study points to targeted policy levers: reducing cross-type transaction costs, counteracting hub concentration through peripheral engagement, and building cross-jurisdictional reputation-sharing infrastructure to support credible bridging.