Success Factors in Civic Crowdfunding in the Hometown Tax Donation System

–Empirical Analysis Using Generalized Additive Model with Machine Learning–

Abstract

This study empirically examines the factors contributing to the success of crowdfunding campaigns organized by local governments under the hometown tax donation system, based on an analysis of 2,533 projects conducted through 2024. The findings from the estimated model confirm that, consistent with previous research, variables such as the target fundraising amount significantly influence the likelihood of successful fundraising. However, in contrast to prior studies, no statistically significant relationship was observed between regional characteristics and fundraising outcomes. Additionally, factors unique to the hometown tax donation system, such as return gifts (given in appreciation of a donation), were found to be significantly associated with fundraising success. This suggests that donors evaluate both the intrinsic content of projects and the distinctive features of the hometown tax system. Moreover, machine learning-based estimations of variable importance reveal that while return gifts contribute to fundraising success, their influence is not overwhelmingly dominant, reflecting a shift in donor preferences.