At the second stage, by using multiple negative binomial regression models, this research constructs three base models and one moderating model to examine the interaction effect of the moderator. The purpose of the three base models is to establish a baseline against the added contribution of the moderator. The first based model examines the relationship between control variables and firm innovation while the second and third base models examine whether absorptive capacity and autonomous R&D climate have the direct impact on firm innovation. The
fourth model examines the interaction effect of the moderator on dependent variable.