Adaptive Capacity and Performance-Based Funding Effectiveness: Evidence from Chinese Vocational Education

YAN HUANG^1*XIN LI²XIAOPING WU¹Yanqiu MAO³JINYU ZHAO²

• ¹Jiangsu Food & Pharmaceutical Science College, Huai'an, China
• ²Jiangsu University, Zhenjiang, China
• ³Fuyang Normal University, Fuyang, China

Employing a mixed-methods research design, the study combines closed-ended Likert surveys (54 items, α=0.78–0.93), semi-structured elite interviews guided by a standardized protocol including open-ended thematic questions, and document analysis (n=1,247) using NLP-based content extraction, this study examines performance-based funding (PBF) effectiveness across 52 Chinese vocational colleges (2018-2023), where the 2022 Vocational Education Law created unique implementation conditions through dual administration (government-led, industry-guided) and provincial funding autonomy. Chinese vocational colleges face distinctive pressures to meet both government performance metrics (graduation rates, skill certifications) and industry employment expectations, with 15-35% of institutional funding linked to outcomes varying by province. Using multi-method causal design integrating staggered-adoption difference-in-differences with Callaway-Sant'Anna estimators, causal forest algorithms for heterogeneous treatment effects, and semi-structured elite interviews (n=67) with document analysis (1,247 records), the analysis reveals substantial heterogeneity in institutional responses within this specific educational context. High-capacity institutions in the sample achieved efficiency gains of 14.7% (95% CI: 11.0-18.4%), while low-capacity institutions showed marginal gains of 1.8% (95% CI: -1.4-5.0%), representing an eightfold differential. This heterogeneity correlates with variations in adaptive governance capacity across four dimensions: institutional memory, structural plasticity, learning orientation, and innovation capacity. Within the studied Chinese vocational colleges, the proposed Adaptive Governance Capacity (AGC) framework demonstrated explanatory power (R²=0.64, p<0.001, n=52) compared to traditional principal-agent models (R²=0.12), requiring external validation. Machine learning algorithms detected potential gaming indicators in 74% of low-capacity institutions compared to 13% of high-capacity institutions, though these patterns require cautious interpretation. These behaviors included grade inflation, selective admissions, and employment data manipulation. Separately, the analysis found associations between PBF implementation and increased institutional inequality, with the Gini coefficient rising from 0.268 to 0.339. However, causality cannot be definitively established given potential confounding factors. Analysis of the studied Chinese vocational colleges suggests that context-specific differentiated approaches warrant consideration: aggressive performance linkage (35-40%) for high-capacity institutions, graduated implementation (20-25%) with targeted support for medium-capacity institutions, and intensive capacity building preceding PBF exposure for low-capacity institutions. These recommendations reflect the particular governance structures and constraints of China's vocational education system and require adaptation for other contexts.