[2Yin-A] Poster 3

[2Yin-A-07] Is Corporate Voluntary Disclosure Nonlinear?

〇Yuichiro Nakai¹, Mitsuo Yoshida¹ (1. Tsukuba Univ)

Keywords:

Voluntary Disclosure,linear model,Threshold

This study moves beyond the conventional assumption of an average positive association in voluntary disclosure research and instead examines the functional form of how disclosure probability responds to different levels of explanatory variables. Using a panel dataset of 5,085 Japanese listed firms (225,738 firm-quarters) from 2009 to 2024, we measure voluntary disclosure by the issuance of corporate activity reports.For three domains—size, experience, and monitoring—we construct decile-based probability curves and classify their response shapes. The results reveal that size and monitoring predominantly exhibit switch-type or mixed patterns characterized by discontinuities or saturation effects, whereas experience mainly follows a gradual, monotonic increase.By visualizing when and how incentives operate across variable levels—patterns that are difficult to capture through average marginal effects—this approach refines both theoretical interpretation and practical reporting strategy. More broadly, the framework provides a non-parametric diagnostic tool for analyzing how binary decisions respond to continuous variables, offering potential applications in artificial intelligence and machine learning research concerned with the structural shape of decision processes.