Abstract: This paper uses stepwise least squares regression, quantile regression, and grey prediction models to form a multi-model integrated method for evaluating the influencing mechanisms and future trends of enterprise research and development (R&D) investment. The results show that the multimodel integrated evaluation method can accurately and intuitively assess the current status of R&D investment in technology enterprises. Regional production levels, government fiscal allocations for science and technology, and enterprise contributions are the three core drivers of R&D funding growth. Government fiscal support demonstrates cross-level sustained effectiveness, with regression coefficients remaining stable in the range of 0.953-1.085, highlighting the stable leveraging role of policy-driven innovation investment. Some enterprises may experience a growth rate adjustment of 3.2%-5.7% over the next five years, signaling structural risks from overreliance on single entities. Based on these findings, the paper proposes the construction of a gradient policy system, enhancing the inclusivity and stability of fiscal support, and fostering collaborative innovation among small, medium, and large enterprises. This study provides theoretical and methodological support for regional innovation policymaking and enterprise R&D investment optimization.
Keywords: enterprise research and development(R&D) investment; least squares regression; quantile
合肥工业大学学报自科版